<%BANNER%>

Paying the Poor

Permanent Link: http://ufdc.ufl.edu/UFE0042984/00001

Material Information

Title: Paying the Poor Bolsa Familia and Child Mortality in Brazil
Physical Description: 1 online resource (116 p.)
Language: english
Creator: GIESE,CLAY W
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: BOLSA -- BRAZIL -- CASH -- CHILD -- CONDITIONAL -- FAMILIA -- MORTALITY -- POVERTY -- TRANSFERS
Latin American Studies -- Dissertations, Academic -- UF
Genre: Latin American Studies thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: PAYING THE POOR: BOLSA FAMILIA AND CHILD MORTALITY IN BRAZIL By Clay William Giese May 2011 Chair: Charles Wood Major: Latin American Studies This thesis examines Bolsa Fam?lia and its effects on child mortality in Brazil. Bolsa Fam?lia is the Brazilian government?s premiere poverty reduction program and it is part of a relatively new family of poverty alleviation programs known as conditional cash transfer programs. The basic premise of the Bolsa Fam?lia program is that the government gives poor families a small sum of money monthly if the family complies with all mandated requirements which include both health and educational requirements. Theoretically, if families are meeting the set health requirements and receiving a small additional monthly salary, we would expect to see a decrease in child mortality. My research hypothesis is that, controlling for several socio-economic factors such as place of residence, region, age, ethnicity, literacy, years of school completed, per capita income and running water in the house, participants in the Bolsa Fam?lia program will be less likely to have a child die than non-program participants. Results of the analysis are conflicting at best and do not offer a clear picture of the program effects. The majority of the results were not statistically significant. However, when separating the results by the region of the respondent (between the High-Mortality region of the North and the Northeast and the Low-Mortality region of the Southeast, South and Central-West) we observe the program participation does not reduce child mortality in the High-Mortality region, but it does reduce child mortality in the Low-Mortality region. This suggests that a better developed healthcare infrastructure (in the Low-Mortality region as compared to the High-Mortality region) may facilitate the program in reducing child mortality.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by CLAY W GIESE.
Thesis: Thesis (M.A.)--University of Florida, 2011.
Local: Adviser: Wood, Charles H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042984:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042984/00001

Material Information

Title: Paying the Poor Bolsa Familia and Child Mortality in Brazil
Physical Description: 1 online resource (116 p.)
Language: english
Creator: GIESE,CLAY W
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: BOLSA -- BRAZIL -- CASH -- CHILD -- CONDITIONAL -- FAMILIA -- MORTALITY -- POVERTY -- TRANSFERS
Latin American Studies -- Dissertations, Academic -- UF
Genre: Latin American Studies thesis, M.A.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: PAYING THE POOR: BOLSA FAMILIA AND CHILD MORTALITY IN BRAZIL By Clay William Giese May 2011 Chair: Charles Wood Major: Latin American Studies This thesis examines Bolsa Fam?lia and its effects on child mortality in Brazil. Bolsa Fam?lia is the Brazilian government?s premiere poverty reduction program and it is part of a relatively new family of poverty alleviation programs known as conditional cash transfer programs. The basic premise of the Bolsa Fam?lia program is that the government gives poor families a small sum of money monthly if the family complies with all mandated requirements which include both health and educational requirements. Theoretically, if families are meeting the set health requirements and receiving a small additional monthly salary, we would expect to see a decrease in child mortality. My research hypothesis is that, controlling for several socio-economic factors such as place of residence, region, age, ethnicity, literacy, years of school completed, per capita income and running water in the house, participants in the Bolsa Fam?lia program will be less likely to have a child die than non-program participants. Results of the analysis are conflicting at best and do not offer a clear picture of the program effects. The majority of the results were not statistically significant. However, when separating the results by the region of the respondent (between the High-Mortality region of the North and the Northeast and the Low-Mortality region of the Southeast, South and Central-West) we observe the program participation does not reduce child mortality in the High-Mortality region, but it does reduce child mortality in the Low-Mortality region. This suggests that a better developed healthcare infrastructure (in the Low-Mortality region as compared to the High-Mortality region) may facilitate the program in reducing child mortality.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by CLAY W GIESE.
Thesis: Thesis (M.A.)--University of Florida, 2011.
Local: Adviser: Wood, Charles H.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042984:00001


This item has the following downloads:


Full Text

PAGE 1

1 PAYING THE POOR: BOLSA FAM ILIA AND CHILD MORTALITY IN BRAZIL By CLAY WILLIAM GIESE A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MAST ER OF ARTS UNIVERSITY OF FLORIDA 2011

PAGE 2

2 20 11 Clay William Giese

PAGE 3

3 To the people of Brazil : It is my hope that the Bolsa Famlia program continues to improve the lives of Brazilians until it is no lon ger necessary to combat poverty

PAGE 4

4 ACKNOWL EDGMENTS I would like to extend a special thanks to everyone that helped me through ou t the process of writing this thesis. Thank you to the Center for Latin American Studies for providing the grant to make my field research possible. Thank you to Ludmila R ibeiro for guiding me and assisting me while I was in Rio de Janeiro. Also, thank you to Dr. Jos Alberto Magno de Carvalho and the entire staff of CEDEPLAR at the Federal University of Minas Gerais for graciously sharing your resources, knowledge and expe riences with me. Thank you to my committee members, Dr. Marianne Schmink and Dr. Renata Serra for their advice and suggestions for improving my thesis. Thank you especially to Dr. Charles Wood for serving as my committee chair and for his tireless work to see thesis project through to its completion. Dr. Wood, your passion and determination truly inspired me when working on my thesis. Finally, I would like to thank my friends and family, especially my parents, for standing by my side and encouraging me the whole way. This would not have been possible without your constant love and support.

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 8 LIST OF ABBREVIATIONS ................................ ................................ ............................. 9 ABSTRACT ................................ ................................ ................................ ................... 10 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 12 2 CONDITIONAL CASH TRANSFER PROGRAMS ................................ .................. 18 What Are Conditiona l Cash Transfer Programs (CCTs)? ................................ ....... 18 History of CCTs ................................ ................................ ................................ ...... 19 Justification for Conditional Cash Transfers ................................ ............................ 22 Brazil and Bolsa Famlia ................................ ................................ ......................... 23 Bolsa Famlia Program Requirements and Conditions ................................ ........... 24 Cadnico ................................ ................................ ................................ ................ 26 Program Benefits ................................ ................................ ................................ .... 28 Program Success ................................ ................................ ................................ ... 29 Criticisms of Bolsa F amlia ................................ ................................ ...................... 30 Political Implications of Bolsa Famlia ................................ ................................ ..... 33 3 WHO GETS BOLSA FAMILIA? ................................ ................................ .............. 36 Distribution of Bolsa Famlia by Region ................................ ................................ .. 37 Bolsa Famlia Distribution by Race ................................ ................................ ......... 39 Bolsa Famlia Distribution by Employment ................................ .............................. 42 Who Qualifies? ................................ ................................ ................................ ....... 43 4 EFFECTS OF CONDITIONAL CASH TRANSFER PROGRAMS ON CHILD HEALTH OUTCOMES ................................ ................................ ............................ 56 Justification for Program Impact on Various Health Indicators ................................ 56 Bolsa Famlia Effects on Vaccination Coverage ................................ ..................... 57 Bolsa Famlia Effects on Pre natal Care ................................ ................................ 60 Bolsa Famlia Effects on Child Weight and Height ................................ .................. 62 Bo lsa Famlia Effects on Child Mortality Rates ................................ ....................... 64 Health Outcomes in Other CCT Programs ................................ .............................. 67 Implications ................................ ................................ ................................ ............. 69

PAGE 6

6 5 CHILD MORTALITY IN BRAZIL ................................ ................................ ............. 71 Determinants of Child Mortality ................................ ................................ ............... 71 Child Mortality in Br azil ................................ ................................ ........................... 73 Recent Child Mortality Statistics ................................ ................................ ............. 75 Leading Causes of Child Mortality in Brazil ................................ ............................. 76 Connecting Bolsa Famlia with Child Mortality ................................ ........................ 77 6 DOES BOLSA FAMILIA REDUCE CHILD MORTALITY? ................................ ....... 84 PNA D 2006 ................................ ................................ ................................ ............. 84 Weighting the Sample ................................ ................................ ............................. 86 Logisitc Regression ................................ ................................ ................................ 87 Hypothe sis ................................ ................................ ................................ .............. 88 Dependent and Independent Variables ................................ ................................ .. 90 Results ................................ ................................ ................................ .................... 91 Focus o n Eligible Families ................................ ................................ ................ 94 Focus on Younger Women ................................ ................................ ............... 94 Focus on Regions ................................ ................................ ............................ 95 Interpretation ................................ ................................ ................................ ........... 99 7 CONCLUSIONS AND FUTURE WORK ................................ ............................... 106 APPENDIX: EXPLANATION OF CREATING A VARIABLE IN PNAD 2006 TO MEASURE THE EXPERIENCE OF CHILD DEATH ................................ ............. 110 LIST OF REFERENCES ................................ ................................ ............................. 112 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 116

PAGE 7

7 LIST OF TABLES Table page 3 1 Average monthly per capita income by region, Brazil 2006 ................................ 49 3 2 Distribution of Bolsa Famlia by Regions and States Brazil 2006 ....................... 50 3 3 Average monthly per capita income by ethnicity, Brazil 2006 ............................. 52 3 4 Distribution of Bolsa Famlia by ethnic groups, Brazil 2006 ................................ 52 3 5 Distribution of Bolsa Famlia among employed and unemployed Brazilians, Brazil 2006 ................................ ................................ ................................ .......... 52 3 6 Bolsa Famlia distribution by monthly per capita income, Brazil 2006 ................ 52 3 7 Distribution of Bolsa Famlia by region within the eligible population (monthly per capita income R$ 140), Brazil 2006 ................................ ........................... 53 3 8 Distribution of Bolsa Famlia by ethnicity within the eligible population (monthly per capita income R$ 140), Brazil 2006 ................................ ............ 53 5 1 Infant mortality rate in Brazil by regions, Brazil 1930 1990 ................................ 80 6 1 Having a child die by running water in the home, Brazil 2006 .......................... 102 6 2 Odds ratio of having a child die by participation in Bolsa Famlia, age, ethnicity, educational attainment, literacy, per capita income, running water, place of residence and region: Brazil 2006 ................................ ....................... 103 6 3 Odds ratio of having a child die by participation in Bolsa Famlia, age, ethnicity, educational attainment, literacy, running water and place of residence: Brazil 2006 ................................ ................................ ...................... 104

PAGE 8

8 LIST OF FIGURES Figure page 3 1 Map of the r egions of Brazil ................................ ................................ ................ 54 3 2 Distribution of Bolsa Famlia within regions, Brazil 2006 ................................ .... 54 3 3 Distribution of Bolsa Famlia by region, Brazil 2006 ................................ ........... 55 5 1 Infant mortality rate and relative variation according to region: Bra zil 1930 1990 ................................ ................................ ................................ ................... 81 5 2 Infant mortality rate (per 1,000 live births): Brazil (1962 2009) ........................... 81 5 3 Under 5 mortality rate (per 1,000) : Brazil (1962 2009) ................................ ....... 82 5 4 Distribution of infant deaths in Brazil (1995 1997) due to perinatal causes ........ 82 5 5 Conceptual Diagra m of the Pathways through which Bolsa Familia Affects Child Mortality ................................ ................................ ................................ ..... 83 6 1 Map of Brazil with c onceptual c omparative r egions of h igh c hild m ortality and l ow c hild m ortality ................................ ................................ ............................. 105

PAGE 9

9 LIST OF ABBREVIATION S BCG Bacillus Calmette Gurin ( a vaccine against tu berculosis) BF Bolsa Famlia Cadnico Cadastro nico CCTs Conditional cash t ransfer programs DPT Diphtheria, Pertussis, Tetanus vaccine IBGE Instituto Brasileiro d e Geografia e Estatstica (the Brazilian Institute of Geography and Statistics) MDS Ministrio do Desenvolvimento Social e Combate Fome (the against Hunger) OLS Ordinary Least Squares regre ssion (a statistical analysis method) Peti Programa de Erradicao do Trabalho Infantil (a cash transfer program of the Brazilian government with the objective of ending child labor ) PNAD Pesquisa Nacional por Amostra de Domiclios (a national household su rvey that is conducted by the Brazilian Geographic and Statistics Institute ) PS D B Partido da Social Democracia Brasileira ( t he Brazilian Social Democracy Party) PT Partido dos Trabalhadores $R The real is the currency in Br azil; 1 R eal = ~ 0.60 US Dollars WWII World War II

PAGE 10

10 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requ irements for the Degree of Maste r of Art s PAYING THE POOR: BOLSA FAM ILIA AND CHILD MORTALITY IN BRAZIL By C lay William Giese May 2011 Chair: Charles Wood Major: Latin American Studies This thesis examines Bolsa Famlia and its effects on child mortality in Brazil. Bolsa Famlia erty reduction program and it is part of a relatively new family of poverty alleviation programs known as conditional cash t ransfer programs. The basic premise of the Bolsa Famlia program is that the government gives poor families a small sum of money mon thly if the family complies with all mandated requirements which include both health and educational requirements. Theoretically, if families are meeting the set health requirements and receiving a small additional monthly salary, we would expect to see a decrease in child mortality. My research hypothesis is that, controlling for several socio economic factors such as place of residence, region, age, ethnicity, literacy, years of school completed, per capita income and running water in the house, participa nts in the Bolsa Famlia program will be less likely to have a child di e than non program participants. Results of the analysis are conflicting at best and do not offer a clear picture of the program effects. The majority of the results were not statistica lly significant. However, when separating the results by the region of the respondent (between the High Mortality region of the North and the Northeast and the Low Mortality region of the Southeast,

PAGE 11

11 South and Central West) we observe the program participat ion does not reduce child mortality in the High Mortality region, but it does reduce child mortality in the Low Mortality region. This suggests that a better developed healthcare infrastructure (in the Low Mortality region as compared to the High Mortality region) may facilit ate the program in reducing child mortality.

PAGE 12

12 CHAPTER 1 INTRODUCTION My interest in Brazil and its struggle to overcome the poverty and inequality that it faces was born out of an experience that happened over 3,000 miles away from the capital of Braslia. I remember the day very distinctly. It was a sweltering hot day in San Salvador, El Salvador and I had just begun my journey of learning about Latin America. I grew up in a small town in North Carolina and traveling to San Salvador r eally opened my eyes to the reality of the world outside of my small town microcosm. When I agreed Just looking around San Salvador I was immediately struck by the fact that El Salvador, (and all of Latin America, as I would later learn), is a place of contradictions where the powerful rich live side by side with the extremely impoverished. The wealthy, upper class families live behind their wrought iron gates, with their persona l security cameras perched sentinel like on top of balconies that afford stunning views of the city below. In stark contrast, homeless children live on the streets of San Salvador. Many people living in rural areas do not have access to electricity, decent education, or sanitary water. One afternoon my group visited the Multiplaza Mall, where a large sign and thought about the homeless men with whom we had just shared a lun ch. I felt disgusted by the thought of families living in the rural communities walking around on their callused feet while a seemingly calloused society pretended not to notice. What more could I want? Fast forward to today. After spending time traveling in Latin America and studyi ng the region for more than five years, I have learned a lot about the inequality that I

PAGE 13

13 experienced for the first time in El Salvador. I learned that Brazil has one of the highest rates of wealth inequality in the world. I lear ned that Rio de Janeiro is similar to San Salvador in terms of contrasting wealth and poverty, except worse. I saw the squalid living conditions that are known as the favelas (slums) of Rio. But, I also learned that the Brazilian government is trying to im plement long term solutions to this problem, including Bolsa Famlia the focus of this investigation. While I was in Rio this past summer, I spo ke to a woman that lives in Roc in h a, the largest favela in Rio, while waiting for the bus one night. We began t o talk about the Bolsa Famlia program which I was researching and what she thought about the program. She emotionally told me how receiving the small transfer of money from the federal government every month allowed her to keep her children in school and provide for the family needs. As we neared my bus stop, she looked me squarely in the eyes and asked me to please write whatever I needed to write to help ensure that the Bolsa Famlia program continues. Though I do not rem ember her name, this is for her Brazil is a country of contradictions. In Brazil, the poor and the wealthy live side by side. Shanty towns, or favelas as Brazilians call them, begin where middle class neighborhoods end. Due to this inequality and the pervasiveness of extreme poverty, th e Brazilian government has implemented several poverty reduction programs in an attempt to alleviate the problems associated with poverty and to reduce the national incidence rate of poverty. One of the largest poverty reduction policy initiatives of the B razilian government is called Bolsa Famlia and it is part of a new family of conditional cash t ransfer programs (or CCTs) that reward poor families by giving them a small sum

PAGE 14

14 of money each month if the family meets certain pre set conditions that the gove rnment establishes to augment human capital. In addition to being gripped by pockets of extreme poverty, Brazil is affected by other social problems, including inadequate educational and health programs. Poor individuals generally (though not always) are m ore associated with low educational levels and poor health outcomes than those that are not poor. Given the idea that all of these problems are related, it is feasible that a program could be designed to counteract the negative consequences of poverty, edu cational and health deficiencies at the same time. That is precisely what a conditional cash t ransfer program, such as Bolsa Famlia is designed to do. The aforementioned conditions that a family must comply with to receive the program benefits are design ed to enhance educational outcomes and to improve the health of program participants. Whether Bolsa Famlia and other CCTs are successful in achieving their desired outcomes is debatable. Some authors, such as Soares et al. (2007: 24), have found positive outcomes, including a reduction of inequality measured by the Gini coefficient, correlated with being a beneficiary of the Bolsa Famlia program. Still others have found little or no evidence of an association between program participation in a CCT and som e of the desired outcomes. For example, Behrman et al. (2009) did not find any discernable relationship between Oportunidades program. High child and infant mortality rates ar e a serious social and health problem in the developing world. Child mortality rates are correlated with availability of safe drinking water, sanitary living conditions and access to healthcare. Improvements in any

PAGE 15

15 of these conditions generally help reduce the child mortality rates for a country or a region within a country. However, they are not the only factors that determine the child mortality rate of a country. A lso highly associated with child mortality rates are household income and poverty levels. O ne study of the relationship between income the infant mortality rate began to rise. In 1971, 1972 and 1973 the infant mortality rate reached a high of just under 95 p er thousand. When the real wage index began to connection is also made clear by the fact that the regions of Brazil with the highest child mortality rates also have the hi ghest poverty rates. It can be argued that variable correlation does not automatically imply causation; however, it should be sufficiently self evident that living in conditions of poverty greatly increases the odds of having a child die at a young age. Bu t how do child mortality rates relate to poverty alleviation programs? I hypothesize that participating in a poverty alleviation program could have positive effects in reducing the likelihood of having a child die at a young age, thereby reducing the child mortality rate. The purpose of this thesis is to explore the possible relationship between participation in the Bolsa Famlia program and child mortality. Because child mortality rates are linked closely to several of the program conditions that are set f orth for participants, I hypothesize that a correlation exists between child mortality and the Bolsa Famlia program. This line of reasoning suggests the following specific expectation: after controlling for many relevant factors including urban residence, level of schooling,

PAGE 16

16 and race among others, program participants are less likely to have a child die than non program participants. Chapter 2 discus ses the rise and importance of c ond itional cash t ransfer programs in Latin America and Brazil due to the pre v ailing poverty and inequality. It begins by addressing inequality in Latin America and outlining the history of CCTs. Then it continues on to discuss the economic justification for using CCTs as a means of combating pov erty and inequality. Chapter 2 conta ins an overview of the specifics of the Bolsa Famlia program and how it operates including program qualifications, requirements and benefits. Chapter 2 concludes by briefly examining program successes and critiques. Chapter 3 examines the targeting of Bol sa Famlia Because it is such a major governmental initiative, it affects millions of Brazilians and in order to evaluate if the program is having the intended effects or not, it is first necessary to know who receives the program benefits. I examine the distribution of program funds according to qualification for program participation, geographic location, race and employment status. Chapter 4 reviews the literature about Bolsa Famlia and its impacts on health outcomes. Programs of a similar nature have demonstrated positive effects on health outcomes in Mexico and other countries. The literature examining how Bolsa Famlia affects health outcomes is scarce and not highly optimistic given the results. Chapter 5 briefly introduces the problem of child mort ality in Brazil. Over the past decades Brazil has made significant advances in combating child mortality. Chapter 5 covers the factors that normally affect child mortality and briefly describes the history of child

PAGE 17

17 mortality in Brazil. Finally, it reviews some of the more recent statistics and research about child mortality and how it is changing in Brazil. Chapter 6 analyzes the PNAD 2006 data to determine if there is a relationship between participation in the Bolsa Famlia program and the likelihood of h aving a child die at a young age. Chapter 6 describes the data set and the methodology used for the study. Furthermore, it interprets the results and offers thoughts on how the Bolsa Famlia program can be improved to better meet its objectives. Finally, Chapter 7 summarizes the results of the study and provides some suggestions for future research.

PAGE 18

18 CHAPTER 2 CONDITIONAL CASH TRA NSFER PROGRAMS What Are Conditional Cash Transfer Programs (CCTs)? Conditional cash t ransfer programs are some of the latest and most innovative tools in the fight against poverty. Implemented and run by national governments these programs give poor families a small sum of money on a regular basis, provided that the families meet certain conditions and comply with the program requ irements. CCTs attempt to identify poor families and provide them with a monetary incentive to meet the program requirements which are set by the government for the benefit of the program participants. Typical examples of program requirements that governm ents employ include educational requirements (such as school attendance and performance) and health requirements (such as receiving proper immunizations and going to health clinics for regular check ups). Brazil is not alone in using conditional cash t rans fer programs to combat poverty. Other countries in Latin America began their own CCTs shortly after witnessing the operating CCTs exist throughout Latin America including Oportunidades Famlias en Accin Juntos Chile Solidario and Bolsa Famlia Oportunidades in Mexico reaches more than 5 million families usehold living in extreme poverty (Levy 2006: 2). Bolsa Famlia on the other hand covers more than 12 million families which is equal to Bolsa 2010). Currently, 29 developing countries have a CCT program in effect, incl uding almost

PAGE 19

19 every country in Latin America 1 and many other countries are planning on developing their own program (Fiszbein et al. 2009: 31). There fore, it is evident that conditional cash t ransfer programs are a major governmental initiative for combati ng poverty in the developing world. Conditional cash t ransfer programs have been hailed by various international organizations, including the World Bank, and development specialists as creative approaches to solving the problem of poverty. Seemingly, much of the excitement and praise for CCTs has been spawned by their rapid expansion and acceptance as a tool to combat poverty. The history of CCT programs is detailed in the next section but suffice it to say that they had humble beginnings. Over the past de cade governments throughout Latin America, and many others across the world, have imp lemented their own versions of conditional cash tr Bolsa Famlia program is unique due to the size and scope of the program, its attempt to deal w ith many different social problems while redistributing wealth and reducing inequality, and its resilience in the face of criticism. H istory of CCTs The mere existence of conditional cash t ransfer programs raises the question, why are they needed in the fi rst place? Poverty has long plagued many nations, Brazil included, and developing countries struggle to alleviate the social ills caused by constant poverty. After WWII, countries actively sought development economists to promote poverty reduction through economic growth. However, they soon realized that economic growth did not necessarily benefit everyone in a country equally. Recently, 1 Uruguay, Venezuela, Haiti and Cuba do not have an operating CCT.

PAGE 20

20 Brazil has achieved success in economic growth, but the growth was not spread equally across all socio economic classes. Therefore, because a significant in equality gap existed before tended to exacerbate the inequality problem rather than solve it. Even today, Brazil still has a very unequal distribution of wealth as evidenced by its Gini coefficient. The Gini coefficient is a measure of wealth equality that varies between 0 (perfect equality) and 1 (perfect inequality). According to the United Nations Development Programme (2009) Human Development Report th e G ini coefficient for Brazil is 0 55 That places Brazil in 15 th place in the region of Latin America, ahead of only Honduras, Bolivia, Colombia and Haiti 2 Perhaps even more staggering, that corresponds to Brazil being ranked 133 rd out of 142 countries i n the world in terms of Gini coefficient. Policy makers acknowledge that CCTs alone will not completely close this wealth gap, nor is that goal even realistic given that exact income equality does not exist anywhere in the world; but the logic of providing additional monthly income is to allow families to buy more nutritious food and meet basic human needs, while simultaneously investing in human capital development. Unfortunately, poverty and inequality in Brazil and all of Latin America was exacerbated by a period of economic history known as the Lost Decade due to economic crisis and subsequent stagnation of the 1970s and 1980s. In response to this period and under considerable pressure from the International Monetary Fund and the World Bank, Latin Americ an countries adopted a neoliberal economic model in the 2 That is if we consider Latin America as the 18 Spanish speaking countries of the Americas, plus Haiti and Brazil. Note that Cuba is not included in this rank because data were not available for Cuba from the 2009 United Nations Development Programme Human Development Report

PAGE 21

21 1990s that emphasized the private sector instead of the state controlled economies. This model was known as the Washington Consensus and at the time, governments in Latin America began to develop a ne w definition of social policy and how it should be approached. Prior to the Washington Consensus, social welfare had traditionally been viewed as the responsibility of the state. However, the post Washington Consensus period saw the development of a new di rection for social policy. According to Maxine Molyneux, a professor of sociology at the University of London, Latin American governments adopted social policies that took the burden and responsibility of social welfare and shared it with the people that t he policies were supposed to be helping. She writes the official policy discourse and forms of entitlement that are being created in Latin America tend to place more emphasis on individual responsibility, while social security is defined in official state ments as no longer residing be interpreted to mean that the individual has to make responsible provision against risks (through education and employment), the family, too, must pl ay its part (through better care), while the market (through private and the voluntary sector) are all involved in the decentering of expectat ions of welfare from the state. (Molyne ux 2006: 430 31) I n short, governments were no longer going to be fully responsible for the implementation of social welfare policy but instead were forcing the population to share in the responsibility of welf are. Thus, new policies such as conditional ca sh t ransfer programs that demanded that participants comply with set requirements made those participants partly in charge of their own development. It was in this context th at CCTs were born. Conditional cash t ransfer programs began around 1997 with Progr esa in Mexico and a trial program in two Brazilian municipalities. Initial indicators of success in both the Mexican and Brazilian programs

PAGE 22

22 helped fuel their expans ransfer program began in 1996 with the inte nt of eradicating child labor. Known as the Program for the Eradication of Infant Work, or Peti ( Programa de Erradicao do Trabalho Infantil ) by its name in Brazil, the program awarded participating families small sums of money to keep their children out of the workforce (Focalizao 2009: 7). CCTs quickly gained popularity, as evidenced by the expansion of the Peti program and the addition of other cash transfer programs. Justification for Conditional Cash Transfers CCTs are designed to reduce current pov erty by providing poor families with additional income while helping reduce future poverty by investing in the human capital of children. Conditional cash transfer programs provide a small sum of money on a regular basis to families that participate in the program in order to provide the incentive for them to meet pre determined conditions. Different CCTs vary in terms of their requirements, but they generally include criteria such as keeping children enrolled in school, vaccinating children and requiring p regnant women to receive pre natal and post natal care. All of the conditions are intended to augment the human capital of the poor population and thereby reduce poverty in the future. Acclaimed economist James Heckman stresses that investing in people whe n outcomes in early childhood have long f 55). Numerous studies document that, by and large, CCTs increase school enrollment among target families and, to a lesser extent, increase use of preventative health

PAGE 23

23 outcomes (such as test scores, child mortality rates, etc.) is mixed at best, and the Oportunidades program (Feiszbein 2009: 127). B razil and Bolsa Famlia By 2003, the Brazilian government operated several different programs, each of which targeted a cause of poverty. These programs included Programa Bolsa Escola (School Scholarship Program), Programa Bolsa Alimentao (Nourishment Fu nd Program), Auxlio Gas (Gas Assistance) and Programa Nacional de Acesso Alimentao (the National Program of Access to Alimentation). Two of these programs, Bolsa Escola and Bolsa Alimentao were transfer programs contingent on stipulated behaviors ( Focalizao 2009: 7). However, there was little coordination between the various programs. federal bureaucracy, each with separate offices, separate administering officials, sepa rate lists of program recipients and separate sources of funding (Soares 2009: 7). Under this system, one poor family could receive benefits from all programs and another poo r family might not receive any benefits (Soares 2009: 7). In 2003, all programs were unified under one program known as Bolsa Famlia except for Peti which was eventually incorporated into Bolsa Famlia at the end of 2005. The unification of all these pr and Combat against Hunger (or MDS by its abbreviation in Portuguese) to end hunger

PAGE 24

24 and poverty in the country. The programs were centralized in order to stream line the administration of the conditional cash t ransfer programs, redefine the eligibility criteria, and organize the data so that families can be served fairly by the program. Bolsa Famlia Program Requirements and Conditions In order to qualify to receive the benefits of Bolsa Famlia the family must have a monthly income less than or equal to 140 reais (Brazilian currency, often abbreviated R$) per family member ( Bolsa 2010). Monthly per capita income is calculated by summing the monthly earnings of each person in the family a nd dividing the total by the number of family members. Furthermore, families with a monthly per capita income between R$ 70 and R$ 140 only qualify for Bolsa Famlia if they have a child that is 17 years old or younger. All families with a monthly per capi ta income less than R$ 70 qualify for Bolsa Famlia regardless of their situation with children ( Bolsa 2010). The gover nment considers these criteria valid indicators of poverty among Brazilian families. In this sense, Bolsa Famlia can be considered uniqu e among CCTs in Latin America because for the poorest of the poor, eligibility depends only on income and not on the presence of children. According to the program website, the conditions are designed to be commitments for both the participating families ( who are expected to actively improve their conditions) and the public services of the government (who are expected to provide access to quality healthcare, education and social assistance). The idea is that children will have to attend school on a regular basis and receive their vaccinations for the family to qualify for the program, and the children will thus be better equipped for the future in terms of both health and educational attainment.

PAGE 25

25 For educational requirements, Bolsa Famlia stipulates that, fo r a family to remain eligible, children aged 5 15 must be enrolled in school and have a minimum attendance rate of 85% (Fiszbein et al. 2009: 45). For children ages 16 17, the attendance rate at school must be a minimum of 75%. The completion of educationa l requirements is monitored by the Ministry of Education. By using the Cadastro nico (discussed below), the government assigns each student a number and a school code based on tion at the municipal level monitors the attendance of all the students whose family receives Bolsa Famlia benefits. According to Soares and Satyro, the Ministry of Education sends a summary of attendance for each municipality to the MDS (2009: 16). In ad dition to educational requirements, participating families must comply with health requirements. Families with children between the ages of 0 and 7 years must follow the recommended vaccination schedule, and should follow the recommended growth and develop ment nutritional accompaniment program (Condicionalidades 2010). Women aged 14 44 that are pregnant must receive pre and post natal care and should also follow the nutritional accompaniment program for babies (Condicionalidades 2010). Completion of health requirements is monitored by the Ministry of Health. Unlike the education r equirements, which are updated every 2 months, health requirements are updated every 6 months (Soares and Satyro 2009: 16). The final requirements are refer red to as Social Assista nce Conditions. If children under the age of 15 are considered to be at risk, or if they are a part of the Program for the Elimination of Child Labor, families must participate in socio educational services provided by the government, with an attendance ra te of at least 85% (Condicionalidades 2010).

PAGE 26

26 The Ministry of Social Development and Combat against Hunger (MDS by its abbreviation in Portuguese) reviews the summaries sent to it by the local municipalities to verify that families comply with all of the re quirements. When a family does not meet the required conditions, the MDS undertakes a series of steps. First, they investigate the reason for not complying with the requi rements. If noncompliance is justifiable, no punitive action is taken. If noncomplianc e is not justifiable, the famil y is given five chances to meet program requirements. At the first offense, the family receives a warning; at the second offense, the family is blocked temporarily from receiving its monthly benefit, but the benefit can be wi thdrawn the next month (Soares and Satyro 2009: 16). The third and fourth offenses result in suspensions of one and two months, respectively, without being able to make up the missed benefit. A fifth means that the family is no longer eligible for support from the program and is replaced by another family (Soares and Satyro 2009: 17). Cadnico Families are selected for participation in Bolsa Famlia based on a proxy means testing system called Cadastro nico para Programas Sociais ( Cadnico ) that identifies poor families. A proxy means test is a statistical analysis of a household data set that is frequently used to identify poor families. Cadnico reports these findings to the Ministry of Social Development and Combat against Hunger, which automatically sel ects which families will participate in Bolsa Famlia Cadnico is a unique means of measuring and registering the poor population of Brazil. Families that have a monthly per capita income of less than half of the minimum wage or less than three times the minimum wage for the entire family are encouraged to enlist in the Cadnico (Cadastro 2010). It was created in 2001, specifically to locate and

PAGE 27

27 register poor families so that they could be enrolled in social programs and receive benefits from the governme nt (Barros et al. 2009: 7). At its inception, local governments hired teams to go door to door and show people how to register in the new system. Teams were sent to the poor city neighborhoods across Brazil, and information detailing the process was dissem inated throughout the country. Beginning in January 2010 families had to register themselves through the Cadnico system and update their information from time to time. According to a news release from the Ministry of Social Development and Hunger Combat, Decree 6.135 of 2007 requires that families re register with Cadnico every other year (MDS 2010). As of March 9 th 2010, only 320,896 families had updated their information in Cadnico That leaves more than 846,000 families that still had to update thei r information before October 31 st 2010 or they would have their Bolsa Famlia program benefits blocked pend ing re registration (MDS 2010). The Cadnico system currently contains information on nearly 19 million Brazilian families, many of which are very p oor and qualify for Bolsa Famlia (Barros et al. 2009: 7). Because families must register themselves and report their own income, some observers question the accuracy of Cadnico for determining which families should receive program benefits. However, many other questions are included in the economic status. This information, coupled with the self reported monthly per capita family income allows the MDS to focus on families who should qualify for Bolsa Famlia and who are not simply under reporting their income. Currently, the MDS makes decisions based on the self identified per capita income. Nevertheless, Barros et al. suggest that the self reported income could

PAGE 28

28 be used in combination with other available informat ion to generate a better predictor of fam ily income (2009: 9). Finally, because it is such a large system for monitoring the poor population of Brazil, Cadnico can have implications for more than just the Bolsa Famlia program. Barros et al. describe Cad nico as a census of th e poor population of Brazil since it contains a wealth of information. They suggest that it could also be used to select participants for other social programs and to measure regional differences in poverty (2009: 10 11). Program Ben efits If all of the conditions are met, the program pays the family a monthly stipend example, extremely poor families (less than R$ 70 per month) receive R$ 68 automatically a child conditional benefit of R$ 22 for each child between 0 15 years old (up to three children), and R$ 33 for each child between 16 17 years old (up to two children) (Benefcios 2009). Families that earn between R$ 70 and R$ 140 are awarded the same amo unt per child, but they do not receive the automatic basic benefit that families in extreme poverty receive (Benefcios 2009). Three types of benefits thus exist: 1) the Basic B enefit of R$ 68, 2) the Variable Benefit of R$ 22 (children ages 0 15, families can receive a maximum of 3 variable benefits), and 3) the Variable Benefit Linked to Adolescence of R$ 33 (children ages 16 17, each family can receive a maximum of 2 variable benefits linked to adolescence). Once chosen to participate in the Bolsa Famli a program, a representative of each family (typically the mother) is given a card, known as the Bolsa Famlia Social Card It is magnetic, personalized, and can be used to withdraw the cash once per

PAGE 29

29 month at designated machines, known as Federal Economic C ashiers Caixas Econmicas Federais (Beneficios 2010). The card also allows holders to access other government programs. Program Success The Bolsa Famlia program is successful in several respects. For starters, it is extremely popular with program part icipants. Participating families are typically happy with program administration and benefits. Marco Weissheimer cites a study in which a group of Bolsa Famlia recipients were interviewed and asked to evaluate the program. Respondents overwhelmingly resp onded that their evaluation of the program was either has a positive evaluation of Bolsa Famlia Only 1.6% of the respondents in the survey In terms of program outcomes, various studies including Weissheimer (2006) and Soares et al. (2007) find that Bolsa Famlia is having a positive effect on the socio economic status of participating families. Weissh eimer cites a survey from the Instituto Datafolha in So Paulo, Brazil that found that from 2003, many voting age individuals had improved their socio economic standing. The s tudy states that from 2003 2006 (the time period during which the vast expansion of the Bolsa Famlia program was carried out by the Lula government), nearly 6 million people advanced from socio economic classes D/E to class C, which is considered middle class (Weissheimer 2006: 104). Although it is not possible to claim that Bolsa Fam lia is entirely responsible for this significant improvement, it is likely that program contributed to the observed improvement.

PAGE 30

30 Similarly, Soares et al. (2007) found that Bolsa Famlia reduced inequality. The authors break income down into various catego ries to analyze the effect of different types of income increases on inequality. Although the amount of money received by a participating family in Bolsa Famlia the distribution of Bosla Famlia funds is of particular benefit to the poorest families, leading the authors to conclude that it is indeed correlated with a decrease in the Gini coefficient in Brazil. Criticisms of Bolsa Famlia In spite of its success and its endorsement by participants, Bols a Famlia has also been criticized. Some of the most common complaints include suggesting that Bolsa Famlia could have ulterior political motives, that it reinforces perpetuated gender stereotypes, and that it weakens the incentive to work. Some scholars claim that Bolsa Famlia is a form o f political clientelism, the use of benefits/favors to gain political support. The manner in which the federal government implemented Bolsa Famlia gave rise to this charge. Marcel Medeiros et al. point to the fact that the federal government initially gave each municipality a quota of people that it could enroll in the program, thereby delegating decision making power to municipal managers who could select their friends, or family members, or people who were politically supporters (2007: 23). Furthermore, they state that Bolsa Famlia contrasts uses a simple per capita income cut off line ( Mendeiros et al 2007: 23) Consequently, when th ere are two families that both qualify for Bolsa Famlia there is no predetermined method for deciding which family should be enrolled first.

PAGE 31

31 Several separate conditional cash t ransfer programs in Brazil were begun under the administration of President Hen rique Cardoso. However, they quickly became associated with President Lula (the succeeding president) and his party the PT ( Partido dos Trabalhadores Bolsa Famlia hermore, the vast majority of the expansion of Bolsa Famlia was authorized by the Lula government so that it grew to include over 12 million families. Even with this expansion, not all poor eligible families receive program benefits and thus the argument for clientelism persists. Additionally, some opponents of the PT claim that Bolsa Famlia is a form of political clientelism that buys votes in the form of social welfare handouts. While there is little doubt that program beneficiaries appreciate the benef its, it is not clear if Bolsa Famlia alone is enough to buy their loyalty and support at the ballot box. Another strong criticism of Bolsa Famlia (and conditional cash t ransfer programs in general) is that it perpetuates negative gender roles in society. Most conditional cash t ransfer programs ( Bosla Famlia included) give the cash transfers to the mothers of the family. Mothers are seen as the homemakers and the ones that will use the money more responsibly to buy food or clothing for their children. One Opportunidades program is to empower poor women by making the cash payments to them. Maxine Molyneux argues that CCTs tend to have the opposite effect. Because women are required to manage the funds and see that their children meet p rogram requirements, the program only solidifies traditional gender roles. Molyneux writes, w ith fathers marginal to childcare and further ma rginalized by the design of the programme, the state plays an active role in re traditionaliz ing gender roles and i In effect, Oportunidades creates a dependency on a

PAGE 32

32 which may enhance their social status and self respect, but nonetheless, in doing little to secure sustainable livelihoods, p uts them at risk of remaining in poverty for the rest of thei r lives. (2006: 440) A third criticism of Bolsa Famlia which applies to almost all social welfare programs, is that cash payments reduce the incentive to work. Brazilians who oppose Bolsa Fam lia claim that some poor people will not look for a job or accept a low paying job because it would mean they would have too high of an income to qualify for Bolsa Famlia However, it is theoretically conceivable that some individuals would think they ca n do better with Bolsa Famlia than with a low paying job. Nonetheless, caution should be employed before assuming that this is a norm for poor families. Clarissa Teixeira measured the effect of Bolsa Famlia on the num ber of hours that people worked She used data from the 2006 Pesquisa Nacional por Amostra de Domiclios ( PNAD ) to compare people who received Bolsa Famlia payme nts with people who did not. According to Teixeira, the results show that Program Bolsa Famlia marginally diminishes the number of hours worked although the effect is n ot uniform among individuals. The impact is more expressive for informal workers, women, low paid workers, and the ones whose wage represents a smaller share of total household income, that is, the other members a part from the household leader. (Teixeira 2009: x xi) B ecause the effect is only observed for some individuals, and because it is small, she concludes that a work reducing effect does not represent a significant threat to the Bolsa Famlia program objectives (T eixeira 2009: xi). Others criticize the effectiveness of educational goals. Program participants may have the incentive to send their children to school, but attendance does not guarantee good performance in the classroom. Similarly, the pr ogram can

PAGE 33

33 require that children go to school, but when the schools are underfunded and of poor quality, attendance does not mean that students receive an education. While I was in Brazil, I asked some people what they thought of the Bolsa Famlia progra m. Of those who were opposed to the program, the most common responses were that it makes people lazy and also that women were having babies just to qualify for program benefits. Quite frankly I was surprised by the last assertion but it should not be writ ten off as out of the realm of possibility. Tina Rosenburg, a journalist for the New York Times, briefly addressed and rejected this criticism arguing that conditional cash t ransfer programs, to the contrary, encourage the use of contraceptives and small encourage larger families in Mexico, for example, three children is the limit. More 2011 ). One study, car ried out by Bruna Atayde Signorini and Bernardo Lanza Queiroz compared data from the 2004 and the 2006 PNAD datasets to see if there was a statistically significant difference in the rate of fertility for program and non program participants, both within a nd between the two years. Signorini and Queiroz found that women who participated in the program actually had somewhat lower fertility rates compared to wo men who did not participate. Although, this study did not support the hypothesis that Bolsa Famlia p romoted higher fertili ty, the authors recognize that more appropriate data and methods could be employed to provide a more robust treatment of the issue. Political Implications of Bolsa Famlia Because it is a social welfare program, the Bolsa Famlia prog ram inevitab ly has far reaching political implications. In the 2006 Brazilian presidential election, the

PAGE 34

34 incumbent president Luiz Incio Lula da Silva ran for re election against opponent Geraldo Alckmin, the candidate for the PSDB ( Partido da Social Democ racia Brasileira election that Lula handily won. Lula did very well in the North and the Northeast regions, which are the two poorest regions of Brazil and therefore the main beneficiaries of the Bolsa Famlia program. What does this imply about the role of Bolsa Famlia in politics in Brazil? First, it demonstrates that social welfare programs can impact the reputation and legacy of politicians. Lula did better in regions wit h higher percentages of program beneficiaries. This could be due to the fact that some individuals aligned themselves with Lula and the PT because they are grateful for the benefits they receive from Bolsa Famlia (Oliveira 2006: 6). However, that is not n ecessarily the case. It could be that the PT and Lula have more progressive social policies that benefit the poor and lower class Brazilians independent of Bolsa Famlia. We cannot conclusively determine the degree to which Bolsa Famlia was responsible fo election, but there appears to be a strong connection between program participation and support for Lula and the PT. The massive program expansion that began in 2003 reinforced and solidified the public association of Lula with the Bolsa Famli a program. Since 2006, the Lula government has further expanded the Bolsa Famlia program. Some intellectuals have taken up the fight to disassociate Bolsa Famlia from Lula. For example, journalist Gilberto Dimenstein argued in the July 3 rd 2006 issue o f Folha de So Paulo that Lula poses a threat to the Bolsa Famlia program. Given the nature of electoral changes, Dimenstein worries the survival of Bolsa Famlia could be threatened if Lula began to

PAGE 35

35 lose popularity, or if he were to lose an election. Bol sa Famlia he contends, should be (Dimenstein 2006). Such fears are not unfounded, but due to a number of converging factors, it seems unlikely that Bolsa Famlia is in any immediate threat of survival. Brazil has enjoyed economic growth and more international recognition throughout the entirety of the Lula administration. In the fall of 2009, Rio de Janeiro was chosen to host the 2016 Summer Olympics, marking the first time the Olympic Games will be held in South America. Brazil also weathered the global financial crisis of 2008 and 2009 better than other countries due to its robust economic growth recently. All of these factors combined to help Dilma Rousseff win the electio n to be the next president of Brazil in picked successor and therefore she represents continuity of PT policies and the insured continuation of Bolsa Famlia Lastly, many Brazilians believe that the Bolsa Famlia progra m is so popular that even opposition parties would not dismantle the program if they were elected to the office of president. Thus, for the time being, it appears that Bolsa Famlia is a reality for Brazilians that is here to stay.

PAGE 36

36 CHAPTER 3 WHO GETS BOLSA FAMILIA? When they began, the programs that make up the current Bolsa Famlia program were scattered, fragmented and less organized than they are now. Even when they were combined to form Bolsa Familia it was still a fledgling program that has since grow n and matured by winning more support from the federal government. I used the PNAD 2006 data set to contrast Brazilians that do and do not receive Bolsa Famlia PNAD is a national survey designed to measure many different indicators of socio economic stat us and general well being. In 2006 the sample included over 400,000 cases, which can be expanded using a weight factor (supplied in the dataset) to estimate the total population of Brazil. Multiplying each case by the weight factor yields a total number of 186,020,850 cases, equal to the entire population of Brazil in 2006 at the time of the survey. Using the PNAD data from 2006, the estimated number of Bolsa Famlia recipients in all of Brazil at that time was 39,107,308 or 21.03% population. B y 2010 Bolsa Famlia had expanded to cover 12 million families (or more than 46 million people) through out the country ( Bolsa 2010). Who are these 12 million families and what do they have in common? To fully understand the range of Braz i lians who participate in the Bolsa Famlia program, it is first necessary to remember the requirements for program eligibility. Bolsa Famlia program requirements state that families that have a monthly per capita income of less than R$ 140 and have childr en between the ages of 0 17 years qualify for the defined as having a monthly per capita income of less than R$ 70, qualify for program benefits regardless of their situation with children. Howe

PAGE 37

37 expected to meet all of the same education and health conditions to continue receiving program benefits. Distribution of Bolsa Famlia by Region Poverty in Brazil is most preva lent in the Northeast. Alt hough the North is largely undeveloped, the Northeast is still the most backward region of the country. In contrast to the cities of So Paulo, Rio de Janeiro and Belo Horizonte in the Southeast which contains the majority of the Brazilian population, the Northeast is known for its poverty. In colonial Brazil the Northeast region flourished on the production of sugar cane for export and a slave labor economy. Once a center of prosperity and economic growth, the Northeast slumped into a decline following t he shift of economic power to the mining region, and the coffee plantations of the Southeast. But with time, the Brazilian economy grew and industrialized to the benefit of So Paulo and the surrounding area much more than the Northeast. Long characterized by racial inequalities, the Northeast found it challenging to escape the legacy of slavery and achieve the level of economic development in the South and the Southeast. The share of national income garnered by each region evidences this inequality. In 197 0, the Northeast contained 30.3% of the population in Brazil, yet only accounted for 12.2% of the national income. The Southeast, on the other hand, accounted for 64.5% of the national income despite having only 42.7% of the population (Wood, Carvalho: 198 8, 72). Complete income equality in Brazil would suggest that a region with 30% of the population would also account for 30% of th e national income. Conditional cash t ransfer programs, such as Bolsa Famlia aim to reduce the poverty rates in the short ter m and to establish the basis for longer term development.

PAGE 38

38 Today, the Northeast has the lowest per capita income in the country and the greatest percentage of population below the pover ty line. Table 3 1 shows the average monthly per capita income for each region of Brazil. The Northeast has an average per capita income of R$ 292.91 and the North region has an average of R$ 327.05. Both of these numbers are low compared to other regions. The Southeast, with an average of R$ 610.31, boasts the highest averag e monthly per capita income in the country. Meanwhile, the South and the Central West regions have average monthly per capita incomes of R$ 590.98 and R$ 558.07 respectively. Therefore, there is a wide gap in the per capita income distribution by region wi th the northern area of Brazil lagging far behind the rest of the country. Thus, it is not surprising to discover that the greatest percentage of Bolsa Famlia recipients lives in the Northeast. Figure 3 2 shows the distribution of Bolsa Famlia within ea ch region. Of the entire Brazilian population, 21.03% participate in the Bolsa Famlia program. This figure visually represents which regions have the highest proportion of their residents enrolled in the program. According to Figure 3 2 40.28% of the peo ple living in the Northeast receive cash transfers from the Bolsa Famlia program. It is clearly visible that the Northeast has the highest concentration of Bolsa Famlia recipients in Brazil. The region with the second highest concentration of recipients is the North region. In the North 25.78% of the population region receives money from Bolsa Famlia The other 3 regions have relatively low concentrations of program participants, compared to the North and the Northeast. In the Central West region 12.70% of the population participates in Bolsa Famlia ; in the Southeast 11.97% participates in Bolsa Famlia ; and in the South, only 11.51% of the population

PAGE 39

39 participates in the Bolsa Famlia program. The observed differences between regions are statistically si gnificant at the 1% level. Table 3 2 further shows that there is also variation of the distribution of Bolsa Famlia within regions. For example, in the North Amazonas and Par account for the population enrolled in Bolsa Famlia There is also significant variation in the Southeast region where Minas Gerais accounts for 11.15% of all Bolsa Famlia recipients. The state of So Paulo appears to account for a large percentage of Bolsa Famlia recipients, but due to its large populati on it has a much lower concentration of Bolsa Famlia recipients than other states Bahia in the Northeast has the greatest number of recipients at 5,404,968 and also one of the highest concentrations in the country of Bolsa Famlia recipients. Figure 3 3 shows all of the Bolsa Famlia recipients and their distribution by region, or in other words, of those that receive Bolsa Famlia where they are located. This figure clearly shows that the Northeast overwhelmingly dominates the Bolsa Famlia program. Of all Bosla Famlia beneficiaries in Brazil, 53.90% reside in the Northeast and 24.00% reside in the Southeast. That means that of all Bolsa Famlia recipients in Brazil (which is 39,107,308), 24.00% live in the Southeast. This may seem like a large percenta lower percentage of program participants per resident of the region than the North. Bolsa Famlia Distribution by Race Because the Northeast has the largest number of poor people in the cou ntry, it stands to reason that the Northeast also contains the largest proportion of Bolsa Famlia participants. The Northeast is also distinguished for having the largest Afro Brazilian population. Given the scholarly interest in racial inequality, it is useful to classify the

PAGE 40

40 percent of Bolsa Famlia participants by race. Brazil is a multi racial country with a history of racial discrimination and thus, there is a significant wealth gap between white Brazilians and Afro Brazilians. Numerous scholars (Neal and Johnson 1996; Altonji, Doraszelski and Segal 1999) have researched causes of the difference in earnings between the two racial groups. According to PNAD 2006 data, on average, w hite Brazilians earn approximately 1.93 times, or roughly $R 316.46 per mo nth per capita, more tha n b lack Brazilians. Furthermore, w hite Brazilians earn 2.21 times or $R 351.19 per capita on average each month than those Brazilians that classified their ethnicity as b rown. These findings are consistent with the existing literatu re on racial income inequality in Brazil and suggest that more Afro Brazilian families should qualify for Bolsa Famlia than white Brazilians. Because each respondent in the PNAD dataset reports his/her per capita income, it is possible to calculate the av erage income for each ethnic group. Granted that this measure is still an estimate and is not precise, because there could be outliers (either extremely high or extremely low values that skew the average so that it does not accurately reflect the distribut ion of the data) that affect the average. However it is highly unlikely that outliers significantly change the average value for any ethnic group due to the extremely large size of the dataset. Regardless of the actual difference in average income, it is c lear that Afro Brazilians earn less than white Brazilians and that the difference is statistically significant at a .000 level of significance. PNAD 2006 data confirm the findings. According to T able 3 4 of all white Brazilians, only 12.89% participate in the Bolsa Famlia program. Because the majority of Brazilians classify themselves as white, that equals approximately 11,868,002 white

PAGE 41

41 recipients of Bolsa Famlia On the other hand, roughly 3,052,882 black Brazilians and 24,017,969 brown Brazilians parti cipate in the program. That is 23.74% of the black population and 30.17% of the brown population of Brazil. Furthermore, 22.94% of indigenous Brazilians receive program benefits. It is evident that the groups that would constitute a traditional definition of Afro Brazilian (black, brown, and indigenous because of the skin color) have a higher percentage of individuals enrolled in Bolsa Famlia than the average percentage of all individuals enrolled in the program (21.03%). If on average, Afro Brazilians rec eive a lower salary than white Brazilians, it stands to reason that Afro Brazilians constitute more of the poor population than white Brazilians. If Bolsa Famlia is properly targeted, it should include more Afro Brazilian participants than white Brazilian s due to the nature of targeting the poorest people in Brazil. Fortunately, Bolsa Famlia appears to target Afro Brazilians effectively by providing them with equal access to the program. When controlling for other confounding variables, Fiszbein et al. co Brazilians were significantly less likely to be excluded [from Bolsa Famlia However, that is not to say that Bolsa Famlia uses perfect targeting techniques and that there is no room for improvement. A s mentioned previously, given two families that qualify for Bolsa Famlia there is no factor for measuring which family is more deserving and should be enrolled first. Is it ideal to enroll more Afro Brazilian families in an attempt to reduce the income i nequality between ethnic groups? The ideal solution would be to enroll the families that are in the most need, regardless of skin color. However, the Brazilian government has not currently developed a means of evaluating families any more in depth than usi ng the per capita income measure and inherently any further

PAGE 42

42 measures taken would draw criticism from one or more segments of the Brazilian population. Bolsa Famlia Distribution by Employment As is common throughout Latin America, Brazil has a high unempl oyment rate. A significant proportion of the Brazilian population is not formally employed but works in the informal sector. Those that work in the informal sector engage in various economic activities such as selling food or other items on the street. Als o, those involved in drug favelas would be included in the informal sector. Thus, the informal sector makes up a large population that might qualify to receive Bolsa Famlia funds because the income is not formally reported to the government and there is no mechanism in place for insuring that families report all of the income earned appropriately in the Cadastro nico The Brazilian government would prefer to have people engaged in the formal economy as opposed to the inform al economy because that way it could collect more taxes. Bolsa Famlia like any other social welfare program has drawn criticism from some opponents for wasting money. Some argue that giving cash to poor families reinforces laziness and discourages the u nemployed and those employed in the informal sector from trying to find a job out of fear that the household per capita income would increase beyond the maximum threshold for receiving program benefits. However, I was not able to find any literature to sub stantiate these claims. Of the people that I spoke to about Bolsa Famlia some repeated these claims about the program contributing to laziness, but others responded that those claims were the standard mantra of the opposition to the ruling political part y the Partido dos Trabalhadores PT

PAGE 43

43 and receiving Bolsa Famlia than being employed and receiving program benefits. According to the PNAD 2006 data, summarized in Ta ble 3 5, of those that are unemployed 20.84% receive benefits from Bolsa Famlia In contrast, for those that are employed, 17.39% receive benefits of Bolsa Famlia That is a relatively small, but statistically significant (at the 1% level) difference due to the large number of available cases. However, this only lends credence to the idea that unemployed individuals are more likely to receive Bolsa Famlia funds than employed individuals, which is the relationship that should be expected because logical r easoning would suggest that being unemployed also leads to a greater probability of having a per capita family income less than R$ 140. Who Qualifies ? Theoretically, all of the recipients of Bolsa Famlia belong to poor families but as I will demonstrate not everyone that is poor receives a share of the funds and sometimes families that should not qualify for Bolsa Famlia are able to exploit the program for their benefit. According to the World Bank Policy Research Report Conditional Cash Transfers: Red ucing Present and Future Poverty 100% of Brazilian families in the bottom quintile of income earnings qualify to receive funds from Bolsa Famlia but only 55% actually receive the funds (Fiszbein 2009: 76). My analysis of the PNAD 20 06 data is summarized in Table 3 6 which corroborates the findings of the research report. Of the individuals who capita income of less than R$ 70 regardless of their situation with children), only 57.62% receive program b enefits. Si milarly, of the individuals who fall in the category of 140 (which is the cut off line for qualifying for the program if the family has children between 0 17 years old), 48.12% receive Bolsa Famlia program benefits.

PAGE 44

44 From these da ta it is clear that not everyone that qualifies for the program is enrolled individuals do not receive program benefits. What are the reasons for these errors of exclusion? Ther e are many possible explanations for this lack of coverage. Fiszbein et al. suggest that due to the limited funds the Bolsa Famlia program is not able to offer all eligible families a place in the program. Since the municipal governments have the authorit y to choose which families eligible families to join Bolsa Famlia and conclude that this is probably the main cause of under coverage (2009: 77). Some people believe that the Bolsa Famlia program is simply not large enough to provide coverage for all of the poor families in Brazil. Soares et al. analyze the expansion of Bolsa Famlia to include more than 11 million families and they determine that latest phases of exp ansion were necessary but not sufficient to Focalizao e Cobertura do Programa Bolsa Famlia: Qual o Significado dos 11 Milhes de Famlias Soares et al. conclude that expanding Bolsa F amlia significantly helped cover more families that qualify for the program, but that it was not enough to reach all eligible families. According to their estimates, the program would have to expand to 14 million families to cover every family living in poverty in Brazil (Focalizao 2009: 23). Another hypothesis about the lack of sufficient coverage for qualifying families is the idea that there is an institutional bias against families that may not be equipped to enroll in the program. Institutional bias could result if families are unable to comply with

PAGE 45

45 Bolsa Famlia conditions. For example, if a rural community does not have a health clinic and the people living there have to travel a long distance to reach the closest health clinic, people cannot comply with the health requirements. Another explanation focuses on the challenges of filling out the Cadastro nico if the family unwittingly makes a mistake in completing t he survey. When this occurs the family might not even be considered by the municipal government to be qualified for the program. Therefore, a mixture of institutional and program factors could pre vent some eligible families from participating in Bolsa Fam lia There is not much research concerning institutional bias and how it can affect/prevent families from participating in Bolsa Famlia and it is an area that merits further study. Finally, poor targeting of program participants is one other commonly cit ed rea son for the lack of coverage of qualifying families. Fiszbein et al. propose that some families to choose not to enroll in the program despite qualifying. People wit h higher income, higher education, or who live in urban areas were less likely to participate than others (Fiszbein 2009: 77). However not all of the exclusion error can be attributed to self selection. When Bolsa Famlia began, program representatives adm inistered the Cadastro nico to determine eligibility for the program. Today, program participants must update their own Cadastro nico in order to prove that they still qualify. This system seems to work fairly well, but it is far from perfect as it opens the possibility of fraud. Somehow families that do not qualify for Bolsa Famlia because they have a per capita monthly income that is too high, are able to receive monthly cash transfers fr om

PAGE 46

46 the program. Table 3 6 indicates that of individuals with a m onthly per capita income of R$ 141 210, 30.98% participate in Bolsa Famlia Seemingly these individuals should not qualify for Bosla Famlia because their monthly per capita income is above R$ 140; however because of the way the question was asked when administering the PNAD survey, income from the Bolsa Famlia program is included in the calculation of per month, they would qualify for and receive Bolsa Famlia and then report their total per capita income to PNAD as being higher than R$ 140. Therefore, this variable exhibits circularity in that given the information as is, we cannot know if receiving Bolsa Famlia benefits is the factor responsible for the per capita inc ome being greater than $R 140. Howev er, somewhat alarmingly, Table 3 6 also shows that individuals in the per capita income categories of R$ 211 280 and R$ 281+ receive Bolsa Famlia 18.08% and 3.73% respectively. The argument of circularity is only val id up to a certain point because the program benefits are not large enough to raise poor families from less than R$ 140 to more than R$ 281 per capita. Therefore, it is probable that some families are committing fraud or that respondents did not truthfully answer all of the questions in the survey. Given the fact that there are Brazilians who are eligible for Bolsa Famlia that are not enrolled in the program, it is interesting to examine who are the individuals that are being excluded from the program. Int erestingly enough, both the region where a person lives and his/her ethnicity significantly affects whether they will be enrolled in Bolsa Famlia or not.

PAGE 47

47 Table 3 7 was generated using the PNAD 2006 data, and it shows a strong bias in favor of the Northeas t and the Central West over other regions. In the Northeast only 37.7% of the people that are eligible for Bolsa Famlia are not enrolled in the program. Likewise, in the Central West 29.8% of the eligible population does not receive Bolsa Famlia Compare d to the other regions, that is a very low percentage. In the other regions of Brazil, the percentage of the eligible population that is not enrolled in the program ranges from 57.0% in the North to 59.9% in the Southeast. That is in accordance with the ge neral sentiment of the Brazilian population. While I was in Rio de Janeiro and I told people that I was studying Bolsa Famlia many of them asked me why I was there instead of in the Northeast. There seems to be a strong bias towards the Northeast because of the perception of it be there is a higher concentration of Bolsa Famlia recipients there than in other regions. By the same token, ethnicity also affects who receives Bolsa Famlia Table 3 8 shows that responden ts who classified themselves as brown were significantly more likely than those of other races to receive Bolsa Famlia if they are eligible. Of the Bolsa Famlia eligible population among brown Brazilians, 45.3% do not participate in the program. Approxim ately 53% of black, white and indigenous Brazilians who are eligible for Bolsa Famlia do not receive program benefits. The ethnic group that appears to be most biased against when eligible for Bolsa Famlia is Asian Brazilians. Perhaps this is due to the fact that Asian Brazilians constitute a small minority of the Brazilian population and there is a perception that Brazilians of Asian descent are relatively well off compared to other ethnic groups.

PAGE 48

48 Table 3 3 shows that on average, Asian Brazilians have a greater monthly per capita income than other ethnic groups. Therefore, if a municipality does have two families that qualify for Bolsa Famlia and one happens to be Asian Brazilian while the other is brown, it is conceivable that the brown family might be chosen for program participation based on the perception that Asian Brazilians are relatively well off. Either way, these findings do not suggest that there is much perceptible racial discrimination against Afro Brazilians because, brown, black and indigen ous Brazilians have just as much or more access to Bolsa Famlia if they are eligible as white Brazilians. Overall, Bolsa Famlia is a fairly well targeted program. The majority of recipients (by percentage) are in the North and the Northeast regions. Bols a Famlia is not large enough however to cover the entire poor population of Brazil. By percentage more Afro Brazilians receive Bolsa Famlia funds than white Brazilians and also, the unemployed are more likely to receive program benefits than those that are employed.

PAGE 49

49 Table 3 1. Average monthly per capita income by r egion, Brazil 2006 Region Per capita income North R$ 327.05 Northeast R$ 292.91 Southeast R$ 610.31 South R$ 590.98 Central West R$ 558.07 Total R$ 490.78 F = 1329035, ss = 0.000 Sou rce: PNAD 2006

PAGE 50

50 Table 3 2. Distribution of Bolsa Famlia by Regions and States Brazil 2006 Within the Region Within the State Region % of BF Recipients # of BF Recipients State % of BF Recipients # of BF Recipients North 9.80% 3,832,516 R ondnia 0.75% 292,765 Acre 0.52% 204,781 Amazonas 2.11% 824,723 Roraima 0.34% 131,406 Par 5.01% 1,960,214 Amap 0.11% 44,687 Tocatins 0.94% 366,486 Northeast 53.90% 21,078,839 Maranho 7.39% 2,890,406 Piau 3.54% 1,382 ,651 Cear 9.35% 3,655,051 Rio Grande do Norte 2.82% 1,101,346 Paraba 4.26% 1,664,055 Pernambuco 7.91% 3,095,041 Alagoas 3.41% 1,331,638 Sergipe 1.45% 568,163 Bahia 13.82% 5,405,968 Source: PNAD 2006

PAGE 51

51 Table 3 2 Continued Within the Region Within the State Region % of BF Recipients # of BF Recipients State % of BF Recipients # of BF Recipients Southeast 24.00% 9,385,754 Minas Gerais 11.15% 4,362,195 Esprito Santo 1.69% 661,876 Rio de Janeiro 2 .54% 992,589 So Paulo 8.60% 3,364,031 South 8.00% 3,128,585 Paran 3.46% 1,351,313 Santa Catarina 0.97% 379,168 Rio Grande do Sul 3.53% 1,381,912 Central West 4.30% 1,681,614 Mato Grosso do Sul 0.73% 283,667 Mato Gr osso 1.13% 440,723 Gois 2.04% 797,905 Distrito Federal 0.44% 172,548 Brazil 100.00% 39,107,308 Total 100.00% 39,107,308 Chi square = 17793294.793, ss = 0.000 Chi square = 20792164.273, ss = 0.000 Source: PNAD 2006

PAGE 52

52 Table 3 3. Aver age monthly per capita income by ethnicity, Brazil 2006 Ethnicity Per Capita Income Income Ratio (White Brazilians: comparison group) White R$ 662.75 1 = (R$ 662.75 / R$ 662.75) Asian R$ 1073.79 0.62 = (R$ 662.75 / R$ 1073.79) Black R$ 346.2 9 1.93 = (R$ 662.75 / R$ 346.29) Brown R$ 311.56 2.21 = (R$ 662.75 / 311.56) Indigenous R$ 353.44 1.88 = (R$ 662.75 / R$ 353.44) Total R$ 490.78 1.35 = (R$ 662.75 / R$ 490.78) F = 2021707, ss = 0.000 Source: PNAD 2006 Table 3 4. Distr ibution of Bolsa Famlia by ethnic g roups, Brazil 2006 From each ethnic group Ethnicity % of BF Recipients # of BF Recipients Population of Brazil White 12.89% 11,868,002 92,071,391 Asian 5.34% 48,278 904,081 Black 23.74% 3,052,882 12,859,656 Brow n 30.17% 24,017,969 79,608,781 Indigenous 22.94% 117,804 513,530 Total 21.03% 39,106,849 185,957,439 F = 2052780, ss = 0.000 Source: PNAD 2006 Table 3 5. Distribution of Bolsa Famlia among employed and u nemployed Brazilians, Brazil 2006 E mployment Status % of BF Recipients # of BF Recipients Working Age Population Employed 17.39% 12,386,995 71,230,563 Unemployed 20.84% 17,499,098 83,968,800 Total 18.98% 29,456,839 155,199,363 F = 299108.3, ss = 0.000 Source: PNAD 2006 Table 3 6. Bolsa Famlia distribution by monthly per capita i ncome, Brazil 2006 Per Capita Income # of BF Recipients % of BF Recipients Brazilian Population R$ 0 70 9,294,055 57.62% 16,129,912 R$ 71 140 14,218,977 48.12% 29,548,996 R$ 141 210 8,318,276 30 .98% 26,850,470 R$ 211 280 3,841,009 18.08% 21,244,521 R$ 281 + 3,438,593 3.73% 92,187,478 Total 39,107,678 21.03% 185,961,377 F = 14580017.292, ss = 0.000 Source: PNAD 2006

PAGE 53

53 Table 3 7. Distribution of Bolsa Famlia by region within the eligible p opulation (monthly per capita i ncome R$ 140), Brazil 2006 Bolsa Famlia Recipient Region Yes No Total North 2,199,879 (43.0%) 2,912,275 (57.0%) 5,112,154 (100.0%) Northeast 14,748,193 (62.3%) 8,927,419 (37.7%) 23,675,612 (100.0%) Southeast 4,376,479 (40.1%) 6,533,269 (59.9%) 10,909,748 (100.0%) South 1,468,797 (41.2%) 2,094,876 (58.8%) 3,563,673 (100.0%) Central West 1,696,869 (70.2%) 720,852 (29.8%) 2,417,721 (100.0%) Total 23,514,200 (51.5%) 22,164,708 (48.5%) 45,678,908 (10 0.0%) C hi square = 17793294.793, SS = 0.000 Source: PNAD 2006 Table 3 8. Distribution of Bolsa Famlia by e thnicity within the eligible p opulation (monthly per capita i ncome R$ 140), Brazil 2006 Bolsa Fam lia Recipient Ethnicity Yes No Total W hite 6,437,393 (46.3%) 7,456,226 (53.7%) 13,893,619 (100.0%) Asian 27,752 (32.4%) 57,934 (67.6%) 85,686 (100.0%) Black 1,767,162 (47.5%) 1,955,602 (52.5%) 3,722,764 (100.0%) Brown 15,207,287 (54.7%) 12,609,195 (45.3%) 27,816,482 (100.0%) Indigenous 73, 549 (46.2%) 85,751 (53.8%) 159,300 (100.0%) Total 23,513,143 (51.5%) 22,164,708 (48.5%) 45,677,851 (100.0%) Chi square = 7863883, SS = 0.000 Source: PNAD 2006

PAGE 54

54 Figure 3 1. Map of the r egions of Brazil Figure 3 2. Distribution of Bolsa Fam lia within regions, Brazil 2006 Source: PNAD 2006

PAGE 55

55 Figure 3 3. Distribution of Bolsa Famlia by region, Brazil 2006 Source: PNAD 2006

PAGE 56

56 C HAPTER 4 EFFECTS OF CONDITIONAL CASH TRA NSFER PROGRAMS ON CH ILD HEALTH OUTCOMES Improving health conditions and partic ularly child health conditions has long been a goal of develop ing countries. Conditional cash t ransfer programs are first and foremost, poverty alleviation programs; however, they have secondary goals of improving health and educational outcomes. Many ques tions rem ain about the effectiveness of conditional cash t ransfer programs. Primarily, are they achieving their stated goals? How much of an impact are these programs having on a range of health indicators including, vaccination rates, pre natal care, maln utrition and mortality rates? Many authors have attempted to answer these questions. Bolsa Famlia includes conditions that directly promote vaccination rates and pre natal care for expecting mothers. Moreover, we can hypothesize that participating in Bols a Famlia might be correlated with lower child mortality rates because the program aims to improve health outcomes. Other conditional cash t ransfer programs across Latin America have had measurable positive influences on the health conditions of program pa rticipants. A review of the studies nonetheless shows how difficult it is to determine that Bolsa Famlia is having a positive impact on health outcomes of program participants. Justification for Program Impact on Various H ealth I ndicators Many politician s focus on GDP growth, poverty reduction, unemployment, inequality and a range of other economic indicators to define development and the success of a program such as Bolsa Famlia I however, would argue that the general wellbeing and health of people sho uld be considered an important facet of measuring

PAGE 57

57 shocking: the squalor, disease, unnecessary deaths, and hopelessness of it all! No man understands if underdevelopment remains for him a mere statistic reflecting low income, cited in Franko 2007: 12). Therefore, health indicators can and should be used as measures of development as well. Furthermore, economic wellbeing is g enerally associated with improved health conditions and if the goal of economic development is to increase wealth and reduce poverty, it follows logically that a result and desired outcome should be better health outcomes. Bolsa Famlia Effects on Vaccinat ion Coverage Participating in a conditional cash t ransfer program such as Bolsa Famlia has the potential to greatly influence health indicators. Vaccinations have a significant effect on child health because they help prevent diseases that can be debilita ting or deadly from a very young age. Jeni Vaitsman and Rmulo Paes Sousa organized a study to evaluate the effects of participating in Bolsa Famlia on health indicators. They used a Propensity Score Matching technique that compares the results of similar families from the treatment group (eligible families enrolled in the program) with the control group (eligible families not enrolled) The paired comparisons were necessary because the families chosen to participate in Bolsa Famlia are not randomly selec ted (they are determined by eligibility). As such, the Vaitsman and Paes Sousa analysis is based on a quasi experimental design. According to Va itsman and Paes Sousa, because the program of vaccination has been a priority of the Ministry of Health and vacc ination coverage in Brazil has grown considerably, there might not be a large difference between homes that have similar conditions of access to pub lic health services. (2007: 32)

PAGE 58

58 But they go on to state that program participation could still have a measur able Bolsa Famlia program can increase vaccine coverage for at least two reasons: first for being a condition of the program, which makes people (especially mothers) worry more about carrying out this type of care; and, secondly through an indirect impact, another way is that it can alter the expectations/behavior of Sousa 2007: 32). Increasing vaccination coverage alone should make for a healthier society and reduce morbidity and mortality rates among children living in conditions of poverty. They attempt to measure program participation by examining an experimental group that receives monthly transfers from the Bosla Famlia program and a control group th at does not. Furthermore, to control for time since entering the program, they calculate the indicators of vaccination considering three different age groups: children 0 6 years old, children 0 2 years old, and children 0 1 year old. By estimating th e program impact for the 0 1 and 0 2 year old age groups, their idea is to control for entrance in t he program inasmuch as younger children have a better chance of having been born while the program was already implemented in their household. Vaitsman and Paes Sousa find that there is not much of a difference in vaccination rates for children enrolled in the Bolsa Famlia program when compared to children that are beneficiaries of some other program. For example, the results on a national level indicat e that children 0 6 years of age that participate in some social assistance program other than Bolsa Famlia are .007% more likely to have up to date vaccinations than those that participate in Bolsa Famlia (Vaitsman & Paes Sousa 2007: 34). But what hap pens if program participants are compared to eligible yet non program

PAGE 59

59 participants? Comparing these two groups, Vaitsman and Paes Sousa find that the differences are again minimal. In the Northeast region, more non program participants were able to present their vaccination card ( 0.050% for 0 6 years at an income cut off of R$200.00), whereas in the South and Southeastern regions, a larger percentage of Bolsa Famlia participants had their vaccination card ready (a difference of 0.067% in favor of progra m participants for 0 6 years old at R$200.00) (Vaitsman & Paes Sousa 2007: 38). Thus, the authors were unable to establish a definitive relationship between program participation and the likelihood of having their children immunized and being able to pre sent a vaccination card. The authors then turn their attention to the required vaccines during the first 6 months of life. In Brazil, every child below the age of 6 months is required to have a BCG (tuberculosis) vaccine, the first and second doses of a nt i polio, DPT (Diphtheria, Pertussis, and Tetanus vaccine) and Hepatitis B vaccines (Vaitsman & Paes Sousa any effect, non program participants are more likely to hav e received all of the obligatory vaccinations. Among children from 0 6 years old (using $R 200.00 as the cut off line for program eligibility), the difference between program participants and non program participants is 0.004 meaning non program partici pants are slightly more likely to have all required vaccinations. The same holds true for children aged 0 2 years old ( 0.013), and 0 1 years old ( 0.047) (Vaitsman & Paes Sousa 2007: 39). This national pattern is observed generally in the Northeast, S outheast and South regions of Brazil, while in the North and Central West program participants tended to have met the obligation of getting vaccinated more frequently than non program participants

PAGE 60

60 (Vaitsman & Paes Sousa 2007: 39). It thus appears that part icipation in Bolsa Famlia has not proved successful in guaranteeing the vaccination of participants any more or less than those that do not participate in the program. Vaitsman and Paes Sousa offer a summary of the finding s and one possible explanation: t he differences in the proportion of vaccinated children are unfavorable for the children that live in treated households [program participant households] in relation to children that live in households that are eligible beneficiaries of other programs an d in relation to children that do not benefit from the program. This pattern repeats itself for Brazil and large regions of the country, with the exception of only the Southeast. One hypothesis that could justify this negative difference in the rate of vac cination is the access to public health services. The beneficiaries of Bolsa Famlia could live in areas of less demographic density and with worse condition s of access to health services. (2007: 40) However, there are many more indicators to examine befor e determining whether Bosla Famlia has had a positive health effect. Bolsa Famlia Effects on Pre natal Care Care for expectant mothers is another central component of child health care. If mothers to be do not receive proper medical attention, their chil riate pre natal care. However, because little research has estimated program effects on pre natal care in Brazil, it is instructive to note a very similar p rogram in Mexico. As in Brazil, the Oportunidades program requires that participants seek appropriate pre natal care as one of the conditions for receiving the cash transfer. Several studies have attempted to determine the effect of this condition. For ex ias (2000) reported an 8 % increase in the number of first time prenatal care visits among first trimester pregnant

PAGE 61

61 Huerta that shows an increase in the percentage of program participants who sought prenatal care. The benefits of receiving prenatal care are multi faceted, and should theoretically provide better care for the mother and child. The same authors of the vaccination report in Brazil, Vaitsman and Paes Sousa, also include a short section about the differences in the realization of pre natal care. They attempted to measure whether or not expectant mothers received adequate pre natal care. Vaitsman and Paes r, the adequacy of pre natal care is a condition of Bolsa Famlia so that it can be expected that pregnant women who receive program benefits have an additional incentive to realize all of the pre natal consultations. Moreover, the perceptions that these women have of the offering of public services can be altered when they begin receiving the (2007: 40). They constructed a dummy variable whereby they assigned a 0 to th ose who did not receive appropriate pre natal care and a 1 to those who did. According to the authors, the minimum number of consultations that a woman should have during gestation is 6 (Vaitsman & Paes Sousa 2007: 40). One of the complicating factors for this study is the fact that only a small number of cases are available for analysis. Of the women ages 10 49 surveyed only 582 or 3% were pregnant at the time of the survey and 101 of these cases had to be excluded because it was not possible to calculat e the indicator of adequate pre natal care for them (Vaitsman & Paes Sous a 2007: 40). In the case of vaccinations, the authors observed little if any difference in receiving pre natal care when comparing recipients of Bolsa Famlia with participants of oth er social programs (Vaitsman & Paes Sousa 2007: 41). When they compared recipients of Bolsa

PAGE 62

62 Famlia to pregnant women who do not receive benefits from any government program, the difference is a 0.748 in favor of Bolsa Famlia recipients with a cut off of R$200.00, but a difference of .0925 with a cut off of $R100.00 (Vaitsman & Paes Sousa 2007: 41). These conflicting results do not allow the researchers to draw firm conclusions about program effectiveness. Seemingly, participating in the Bolsa Famlia pro gram does not increase the likelihood of pre natal care. A more positive assessment of these findings concludes that the absence of a program effect is due to the fact that the rate of pre natal care is similar for all groups, which itself is an accomplis hment considering that the families in the Bolsa Famlia program are among the poorest and historically least served families in Brazil. Bolsa Famlia Effects on Child Weight and Height In addition to vaccination rates and pre natal care, program participa tion could also have an impact on indicators of child malnutrition, as indicated by average child weight and average child height. In a study published in the Journal of Nutrition Morris by factors that include lack of access to nutritionally rich diets (1), inadequate infant feeding practices (2), and repeated illness (3). All of these factors are related to poverty, with the result that, within countries, stunting consistently affects c hildren from poorer families more than those a beneficiary of Bolsa Famlia on child height and weight. At the time of the study, the Bolsa Famlia program as it is toda y did not exist. Before Bolsa Famlia existed there were var ious other conditional cash t ransfer programs that were targeted towards specific needs. For example Bolsa Escola included educational requirements and Bolsa

PAGE 63

63 Alimentao included nutritional and h ealth requirements. The Bolsa Alimentao recipients were the target study group of this analysis. beneficiaries with matched individuals from households that were originally selec ted to receive the benefit but who subsequently were excluded due to 1 of 3 quasi random, factors, such as receiving benefits from another program like Bolsa Escola that could Bolsa Escola beneficiary status and Bolsa Alimentao beneficiary households had significantly lower weight f or age at the al. 2004: 2339). This would seem to indicate that receiving benefits from Bolsa Alimentao does not significantly improve the average weight for age of children living analysis of the routinely recorded weight data indicated that there was weak or no evidence of a weight difference between Bolsa Alimentao beneficiary child ren and excluded children at the time of enrollment (P=0.063). However, every additional month of receipt of Bolsa Alimentao transfers was associated with 31 7 g rams (SE) less weight gained (P<0.001). Over a 6 month period, this implies tha t Bolsa Alimentao beneficiary children gained 183 g rams less than excluded children of the same ages. If the same analysis is repeated excluding all children in households receiving Bolsa Escola benefits, an even larger differential growth rate of Bolsa Alimentao beneficiary children is found, with each additional month of receipt of Bolsa Alimentao transfers

PAGE 64

64 associated with 40 9 g rams 2339). Similar results were found in studies o f program impact on average child height for age measure s Morris et al., showed negative outcomes for the coefficient of the impact on height for age: 0.110 for children younger than 24 months, 0.190 for children from 24 48 months and 0.040 for childre n from 49 83 months (Fiszbein et al. 2009: 146). Again, this signifies that participating in the program is correlated with a more adverse outcome. The authors conclude that although being a program beneficiary tends to increase the availability of nutriti ous foods for a household, there might still be a negative impact because of the incentive effects. That is to say that, mothers might have believed that receiving the benefits of Bolsa Famlia were also contingent on the ir children being underweight, e ven though the transfers depend solely Such a perspective could exist as Morris et al. indicate, since a subsidized nutritional supplement program did function on the basis of having underweight children in the past. Specifically, the Incentivo para o Combate de Carncias Nutricionais program provided mothers with powdered milk if they had underweight children (2004: 2340). This could be one possible explanation, but we have no way of verifying the validity of this cla im. Bolsa Famlia Effects on Child Mortality Rates Child healthcare can be measured in a number of different ways. Depending on the gravity of the health situation, some indicators might offer a more accurate picture of the panorama of the health problems facing a country than others. For example, a country with good health conditions might be more concerned with average child weight and height indicators, whereas countries with poor health conditions might be focused

PAGE 65

65 on child mortality rates and vaccinatio n coverage. Brazil is a middle income country with relatively good health indicators, but it still faces the challenge of improving its infant mortality rate which is 17 deaths per 1,000 live births, according to the World Bank. Reducing poverty is a bigg er concern for Brazil, but as Sonia Rocha points out, poverty is not synonymous with a range of problems that are commonly associ ated with poverty. She writes, m arginality, malnutrition, unemployment, illiteracy are all critical problems in Brazil, but the y are not synonyms with poverty, nor are they always associated with it, especially if poverty is defined as a lack of income, and not as a symptom of diverse deficiencies. (Rocha 2008: 103) fining a poor sub population and from there inferring that the poor are malnourished is a grave conceptual a person malnourished, nor does being malnourished make a pers on poor. However, poverty and malnourishment are indeed positively correlated. In their study of demography and inequality in Brazil, Charles Wood and Jos Alberto Magno de Carvalho (1988) document a correlation between income and the infant mortality rate They analyzed data from So Paulo from 1963 1979 and found that as the real wage index increased, infant mortality rates declined. Then during a period of economic hardship and declining real wages, the infant mortality index increased. Referring to a gr aph that shows the relationship betwee n the two measures, they write t he trends are nearly mirror images of one another, indicating a strong inverse relationship between the two. The decline in the real value of the minimum wage from 1964 through the early 1970s is associated with a rise in infant mortality. The upward trend in the death rate, however, was reversed in the late 1970s, when the minimum wage regained purchasing power. (Wood a nd Magno de Carvalho 1988: 116)

PAGE 66

66 Therefore, it is evident that there i s an inverse relationship between income and infant mortality rates. Conceptually, that implies that changes in the real income of a family could affect the child mortality rates. Thus, it is both possible and logical to hypothesize that receiving a monthl y cash stipend from a conditional cash t ransfer program could have an effect on the child mortality rates of program participants. Although it is theoretically possible to foresee a relationship between poverty reduction and child mortality rates, it woul d also be prudent not to have expectations that are too high. First of all, it should be noted that the death of a child is an extreme and relatively rare event, at least compared to child morbidity. Many children get sick; only a few of the m actually die, even in high mortality regimes. Second, a review of the literature reveals very weak relationships between participation in the Bolsa Famlia program and a range of health indicators including vaccination rate, pre natal care and child height. Sonia Rocha writes d mortality rates due to progresses in medicine, in particular preventative measures such as vaccination coverage, and in the infrastructure of basic sanitation, not depending, the refore, strictly on socio economic improvements as measu res of the income. (2008: 106) What then are the factors that affect child mortality rates? Vaccination coverage, access to health clinics, proper nutrition and safe drinking water are all factors tha t the government can improve for the people without necessarily improving their situation of poverty (as defined by insufficient per capita income). Rocha goes on to analyze the c orrelation between the reason for mortality and extreme poverty (assuming tha t extreme poverty would be correlated more with mortality tha n slight poverty). She states,

PAGE 67

67 o bserve that the correlation for the country as a whole is elevated, incidentally as was expected, but the result of 0.71 is far from indicating that the proportion is an excellent proxy for infant mortality. The results by region provide evidence that, the same in the Northeast and the North, the poorest regions and where infant mortality is also more elevated, there exists a great diversity of explicative situation s for infant mortality, which results in relatively weak correla tions between the two variables. (Rocha 2008: 118) Thus, I would caution that the results of a study of the relationship between participation in the Bolsa Famlia program and child mortality rates are not likely to provide a definitive answer to the relationship between poverty reduction and child mortality. Studies should also consider the strong impact that public health policies have made in reducing the child mortality rate. I was not able to find any studies that specifically attempt to analyze the effects of program participation on child mortality rates. Health Outcomes in Other CCT Programs As previously mentioned, other c ountries in Latin America have conditional cash t ransfer programs that have had positive effects on the health outcomes of program participants. Two such programs are the Famlias en Accin program in Colombia and the Oportunidades program in Mexico. CCTs are effective at increasing the use of preventative healthcare se rvices. The authors of a World Bank policy research report positive impact on the health center visits by children aged 0 1 (Fiszbein et al. 2009: 137). Similarly, the impac t for children aged 2 4 was 33.2% and both w ere significant at the 1 % level (Fiszbein et al. 2009: 137). Almost every other Latin American country included in the report exhibited the same positive correlation between program participation and the n umber of visits to health centers. Notably, only Mexico exhibited a slightly negative correlation for children from 0 2 years old (Fiszbein et al. 2009: 138).

PAGE 68

68 Other studies contradict the findings in Mexico. Gertler (2000) documents a significant increase in the demand for preventative healthcare services among program participants in rural areas (Levy 2006: 49). The sam e study foun d a 30 to 60 % increase Opportunit ies program (Levy 2006: 49). However, simply visiting a health center does not result in better health outcomes for program participants. Women who participate in the program are required to visit a health center, but visits alone do not improve their heal th. What matters are the lectures participants are required to attend, which encourage behaviors that improv e their health in the long run. Studies reach various conclusions about whether participation in a CCT program is associated with taller children. For example, numerous studies (including Gertler 2004) found a generally positive impact of program participation for child height in Mexico (Fiszb ein et al. 2009: 147). Attanasio et al. (2005), also found a positive impact for program participation 146). However, Morris, Olinto et al. (2004) found a negative effect among children in Brazil participating in the Bolsa Alimentao program (Fiszbein et al. 2009: 146). The conflicting resul ts indicate that, with respect to the height for age indicator, the findings are inconclusive. Oportunidades program, receiving appropriate pre natal care is one of the conditions for receiving the cash transfer. It serves to remember that Sk oufias (2000) and Hernn dez and Huerta (2000) found positive correlations between program participation and pre natal care in Mexico. Also, u sing municipal administrative data, B arham (2005) found an 11 % decrease of infant mortality rates among Opor tunidades

PAGE 69

69 participants (Levy 2006: 51). Overall, the studies cited seem to conclude that participating in the Opportunities program in Mexico has positive health benefits in terms of prenatal care and infant mortality rates. Prenatal care is especially imp ortant because pregnant women are more vulnerable to health complications than other individuals, and a lack of proper care could result in problems for the development of the child or a failure to detect potential problems. Because of conditional cash t r ansfer programs focus on developing human capital for young members of society, many studies address the health effects for children in participating families. Morbidity, or the incidence rate of disease, is an important indicator for determining the prev alence of disease and determining the health of a community. Paul Gertler (2000) determined that children who participated in Oportunidades developed more disease resistance than children who did not participate in the program. These findings are reflected disease among children aged zero to two years dropped 12% compared with the incidence among non progr am children; that figure was 11% for children aged three to contribute to an effective explanation of these findings. First, program participants might have a more nutritionally balanced diet regular check ups that should be su ccessful in preventing more disease. Later studies, specifically Bautista and others (2004), substantiate these results. Implications The existing literature and studies show a significant increase in the probability that pare nts who participate in certain conditional cash t ransfer programs use educational and h ealthcare services.

PAGE 70

70 suggest that program participation is responsible for a decrease in infant mortality rates, decreased child morbidity rates, increased ch ild height for age measurements, and increased prenatal care. These results are certainly encouraging that CCTs are able to achieve their stated goals, at least to some degree. However, evaluations of Bolsa Famlia do not show the same encouraging results. Bolsa Famlia does not appear to be associated with any positive outcomes on child health indicators thus far studied. Where does that leave us in relation to the overall status of Bolsa Famlia as a policy that is supposed to improve child health indicat ors? First of all, it is necessary to remember that Bolsa Famlia is designed primarily as a poverty alleviation program and secondarily as a human capital development program. Yes, better child health outcomes are certainly desired outcomes, but not findi ng the results that we would like does not indicate that the program is a total failure. There are still many areas of health indicators to explore in connection to Bolsa Famlia For example, does program participation increase usage of public health serv ices or visits to health clinics? Also, are there institutional factors, such as limited access to health clinics that impede or make it difficult to comply with program requirements? Negative findings do not mean that we should simply write the program of differences of implementation could benefit policy makers in Brazil and allow them to make some changes to help Bolsa Famlia comply with its stated objectives.

PAGE 71

71 C HAPTER 5 CHILD MORTALITY IN BRAZIL Determ inants of Child Mortality For the purposes of mortality studies, demographers typically define the child maturation process as the years from birth to age 5. This is because the majority of child deaths occur within the first year of life and by the age o f 5, mortality rates typically are closer to the rate of the 5 9 age category (Chen 1983: 204 205). There are These factors can be categorized into four main groups of the proximate determinants of child mortality, including 1) parental factors, 2) nutrition and diet, 3) infections and infestations, and 4) childcare factors (Chen 1983: 200). Proximate determinants are variables that have a direct impact on the mortality of a child. More specifically, infections, protein survival. The key insight is the notion that, as the term indicates, a proximate determinant is a variable that is most closely associated (in a causal sense) with the probability of death. This means that other variables that can influence child mortality must necessarily operate through one or more of the proximate determinants. Common sense can tell us reflection is sufficient to recognize that income per se has no direct e ffect on child contribution was to identify a finite and exhaustive set of such proximate determinants.

PAGE 72

72 Poverty is therefore a factor that indirectly has an effect on child mortality by having an impact, to one degree or another, on almost all of the proximate determinants. Other indirect variables are socio economic variables, such as education level, which could affect the knowledge a female has about how to best care for her newborn child. However, a mother not knowing about the importance of a healthy maternal diet does not directly predispose her child to health problems at an early age. Socio economic factors that impact child mortality operate through proximate determi nants. Therefore, when analyzing the data from Brazil, I will attempt to control for socio economic variables, as much as possible given the constraints of the dataset. Theoretically, if one were able to control for all conceivable indirect variables, any observed effect from participating in the program would have a direct impact on the proximate determinants. With respect to the proximate determinants of child mortality, it is necessary to note a few of the factors that lead to a higher incidence of child mortality. Parental that affect the child; typically the child is woman having a child at very young (under 17) or very old (over 35) ages leads to a greater risk of child mortality (1983: 208). For this reason, in the analysis of the PNAD data, I will restrict the sample size to women aged 18 34, bearing in mind that older women are more likely to have children that were born and matured to the age of 5 before the family was ever incorporated into the Bolsa Famlia program. Other parental factors that greatly impact child mortality are premature births and maternal nutritional status. If the mother does not consume sufficient n utrients for herself and the baby while

PAGE 73

73 she is pregnant, the child will be born underweight which also raises the risk of child mortality. The fact that nutritional and diet factors are proximate determi nants of child mortality seems self evident. If a ch ild does not receive proper nutrition, he/she could become malnourished which in turn leaves the body weak and sometimes unable to defend against infections. Infections and infestations constitute a third group of proximate determinants. According to Chen childhood deaths are diarrheal diseases (due to shigella and E. coli bacteria and rotavirus); respiratory diseases (such as pertussis, diphtheria, tuberculosis, and bacterial and viral pneumonias); and other infectio Children can pick up infections in numerous different places either through direct contact with bacteria, airborne exposure or consumption of contaminated food and water (Chen 1983: 213). The final group of proximate dete rminants of child mortality is childcare factors. These include vaccinations, access to medicine and healthcare. Childcare factors are thus easily impacted by the development of new technology/medicines which has played a critical role in reducing child m ortality rates across the globe. Child Mortality in Brazil Brazil has made advances in reducing child mortality, especially in the post WWII period. Due in part to access to better technology and better understanding of the diseases that affect children, B razil and other developing countries were able to catch up to more developed countries and drastically reduce their child mortality rates.

PAGE 74

74 Eduardo Arriaga is quoted in a study of the Brazilian Institute of Geography and Statistics, saying t hese countries ( including Brazil) do not have to develop and maintain a major medical establishment of their own; rather they can import new techniques, discoveries, or drugs from more advanced nations, as well as receive international financial and material aid. Hence th e public health deal on medical progress and development in other countries. In other words, assuming that most new medical discoveries occur in the most advanced countries, the pu blic health program of an underdeveloped country is related more to the economy of the advanced countries than it is to it s own economy. (Simes 1999: 14) However, not all authors agree that developing countries can simply reduce the child mortality rate b y importing technology and new medicine. Some authors, including Susan Scrimshaw (1974), argue that it is really better nutrition that causes a drop in the child mortality rate. Either way, the post war period of Brazilian history witnessed a dramatic drop in child mortality rates. From 1945 to 1990, the infant mortality rate for all of Brazil fell from 144 deaths per 1,000 children to 48.3 deaths per 1,000 chi ldren (Simes 1999: 20). Table 5 1 was composed using census information and results from the PNAD Observing Table 5 1 shows sustained and significant decreases in the infant mortality rate in each region of Brazil. Some regions made more progress than others during the period from 1930 1990. For example, the North and the Northeast regions each beg an at approximately 193 deaths per 1,000 children in 1930 but the North ended with 44.6 deaths per 1,000 children whereas the Northeast had 74.3 deaths per 1,000 children in 1990 (Simes 1999: 20). What could have caused the Northeast to lag behind other r egions in controlling the infant mortality rate? Could it be due to the pervasivene ss of poverty in the Northeast?

PAGE 75

75 Figure 5 1 shows the same information but in a different format. Figure 5 1 visually highlights the dramatic advances Brazil achieved in re ducing infant mortality rate after WWII. As of 1990, the Northeast was registering the highest infant mortality rate with 74.3 deaths per 1,000 children and the South was registering the lowest infant mortality rate with 27.4 deaths per 1,000 children. Rec ent Child Mortality St atistics Since that time, Brazil has continued to make advances in reducing the child mortality rate. The World Bank releases data known as the World Development Indicators every year to track the progress of developing countries in c ombating the problems that they face. According to the s e data, Brazil has been providing more of its citizens with access to sanitation facilities and has also witnessed a decline in malnutrition (World Bank Online). Overall, this has helped contribute to a dramatic decline in the infant mortality rate and the under 5 mortality rate. As shown in Figure 5 2 t he infant mortality rate has fallen over the past decades and is now approximately 17 deaths per 1,000 live birt hs (World Bank Online). Figure 5 2 show s the sustained decline of the infant mortality rate even into the new millennium. Similarly, the mortality rate for children under the age of 5 has also been reduced drastically. The under 5 mortality rate is approximately 20.6 deaths per 1,000 childr en ( World Bank Online). Figure 5 3 highlights these findings. Overall, Brazil has made tremendous progress in combating infant and child mortality. Nonetheless, there is still room for improvement and programs such as Bolsa Famlia have the potential to help reduce the infant and child mortality rates even further.

PAGE 76

76 Leading Causes of Child Mortality in Brazil Over time, the leading causes of death in Brazil have changed. While diarrhea and respiratory infections used to account for the largest percentage of ch ild deaths in Brazil, perinatal conditions now account for more than half of all child deaths. According to Cesar Gomes Victoria and Fernando Celso Barros, perinatal conditions accounted for 56.8% of all infant deaths in Brazil from 1995 1997 (2001 : 36 ). T he perinatal period refers to the period of time from the 2 2 nd week of gestation to the 7 th completed day after birth. originates during the perinatal period and leads to child death after birth. Perinatal causes can be further subdivided according to the problem that occurs perinatal cause infant deaths in 1996 7, 60.7% were due to respirator y or cardiovascular conditions specific to the perinatal period, 8.4% were due to problems affecting fetal growth and/or duration of the pregnancy, 6.9% due to perinatal problems related to pregnancy complications, 0.3% due to birth trauma and 23.7% due to the infant deaths due to perinatal causes are due to respiratory distress syndrome, 11% are due to hypoxia or anoxia and 28.7% are due to other respiratory conditions (Gomes Victoria 2001: 37). Together, these respiratory and cardiovascular conditions account for approximately 3/5 of all infant deaths due to perinatal causes. Figure 5 4 shows the distribution of infant deaths in Brazil due to perinatal causes. The secon d leading cause of infant death in Brazil is congenital malformations. These malformations include physical defects present at birth due to genetics, exposure of the fetus to toxic substances (including drugs and alcohol), and other unknown

PAGE 77

77 reasons. Malfor mations accounted for 11.2% of all infant deaths in Brazil from 1995 1997 (Gomes Victoria 2001 : 36 ). Examples of malformations include heart defects, Down syndrome and spina bifida. Congenital malformations are very difficult to prevent. The use of vaccine s and medicines makes it easier to reduce the number of deaths due to infections and disease. Therefore as the number of deaths due to infections and disease decreases, the relative percentage of deaths due to malformations increases. Acute respiratory ill nesses are t he third leading cause of infant death in Brazil. The majority of these deaths are caused by pneumonia. Finally, diarrhea and other infections are the fourth leading cause of infant mortality in Brazil (Gomes Victoria 2001: 36) Connecting Bols a Famlia with Child M ortality Theoretically, it is logical to expect a correlation between Bolsa Famlia and child mortality rates because several of the factors that influence mortality rates should be impacted by program participation. Most importantly, children could potentially die from a number of different diseases. Bolsa Famlia requires participating families to follow the standard vaccination schedule for their children which should help protect them against these diseases. Furthermore, Bolsa Fam lia participants should follow the guide for growth and development. To do so requires regular visits to a local health clinic which should help keep children healthy and increase the knowledge of the availability public health services for participating f amilies. In the event of an emergency, Bolsa Famlia participant families should at least be more familiar with getting help from the local susceptibility to disea se. According to a study by Marco Aurlio Weissheimer, 76.4% of the families interviewed spent the money received from the Bolsa Famlia program on

PAGE 78

78 food for the family (2006: 94). Also encouragingly, 11.1% of the families bought school supplies and 5.4% bo ught clothes (Weissheimer 2006: 94). Furthermore, over 80% of the families studied in the Bolsa Famlia program responded that they ran out of food before they had the financial means to buy more food. Therefore, it is clear that Bolsa Famlia is very impo rtant for families to buy the food they need to survive and stay healthy. Food insecurity is still a problem in Brazil because extremely poor families cannot guarantee that they will always have enough food to eat. The Brazilian Institute of Geography and Statistics uses a scale to measure food insecurity that includes four categories: nutritional security, slight nutritional insecurity, moderate nutritional insecurity and severe nutritional insecurity (Weissheimer 2006: 43). Weissheimer states that of the participating families that were interviewed, before entering the Bolsa Famlia program, 58.3% reported that in the past three months, a family member had to forgo eating or eat less due to a lack of food (2006: 96). This represents a situation of at leas t moderate nutritional insecurity. However, after joining Bolsa Famlia the percentage was reduced to 48%, representing a more stable food situation for many families (Weissheimer 2006: 96). Further data to support the better nutritional outcome for progr am partic ipants includes 85.6% reporting an improvement in alimentation after joining the program, 59.2% reporting an increase in the quantity of food consumed and 73.3% citing an increase in the variety of food (Weissheimer 2006: 96). Figure 5 5 shows a conceptual framework of how Bolsa Famlia indirectly impact s child mortality through t he proximate determinants of parent al factors, nutrition and diet, infections and infestations, and childcare factors. Although there are many

PAGE 79

79 causes of child mortality, some are easier to solve than others. Take for example the proximate determinants of child mortality: parental factors, nutrition and diet, infections and infestations and childcare factors. Conceptually we can also group the determinants of child mortalit y in two groups: exogenous and endogenous factors. Exogenous factors are external factors that cause child mortality including nutrition and infection or infestation. Endogenous factors, on the other hand, are internal factors such as parental factors or b irth trauma. This distinction is important because it is easier to reduce child mortality due to exogenous factors. Over the past decades, Brazil has effectively reduced the child mortality rate by providing better vaccination coverage and ensuring that ch ildren have access to better food. The fact that perinatal causes and malformations have replaced diarrhea and infections as the leading causes of child death evidences the fact that the Brazil has been successful in combating exogenous causes of child dea th. However, it remains to be seen if in recent years, Bolsa Famlia has contributed to a r proximate determinants of child mortality and the potential for Bolsa Famlia to affect both endogenous an d exogenous causes of child death, Chapter 6 analyzes the effect of Bolsa Famlia on the child mortality rate in Brazil.

PAGE 80

80 Table 5 1. Infant mortality rate in Brazil by r egions, Brazil 1930 1990 Year Brazil North Northeast Sout heast South Central West 1930 162.4 193.3 193.2 153.0 121.0 146.0 1935 152.7 170.0 188.0 145.0 120.0 133.0 1940 150.0 166.0 187.0 140.0 118.0 133.0 1945 144.0 156.0 185.0 130.0 113.0 123.0 1950 135.0 145.4 175.0 122.0 109.0 119.0 1955 128.2 127.5 169 .6 108.0 94.7 114.0 1960 124.0 122.9 164.1 110.0 96.0 115.0 1965 116.0 111.3 153.5 96.0 84.0 99.0 1970 115.0 104.3 146.4 96.2 81.9 89.7 1975 100.0 94.0 128.0 86.0 72.0 77.0 1980 82.8 79.4 117.6 57.0 58.9 69.6 1985 62.9 60.8 93.6 42.6 39.5 47.1 1990 48.3 44.6 74.3 33.6 27.4 31.2 Sources: Demographic Censuses 1940 1991. Rio de Janeiro: IBGE, 1950 1997; Pesquisa nacional por amostra de domiclios (PNAD) 1992 1993, 1995. Rio de Janeiro: IBGE, v. 15 17, 1997. (Simes 1999: 20)

PAGE 81

81 Figure 5 1. Infa nt mortality rate and relative variation according to region: Brazil 1930 1990 Sources: Demographic Census 1940 1991. Rio de Janeiro: IBGE, 1950 1997; Pesquisa nacional por amostra de domicilios (PNAD) 1992 1993, 1995. Rio de Janeiro: IBGE, v. 15 17, 1997. (Simes 1999: 21) Figure 5 2. Infant mortality r ate (per 1,000 live births): Brazil (1962 2009)

PAGE 82

82 Source: Source: World Bank Online. http://data.worldbank.org/indicator/SP.DYN.IMRT.IN/countries/BR?display=graph Figure 5 3. Under 5 mortality r at e (per 1,000): Brazil (1962 2009) Source: World Bank Online. http://data.worldbank.org/indicator/SH.DYN.MORT/countries/BR?display=graph Figure 5 4. Distribution of infant deaths in Brazil (1995 1997) due to perinatal causes

PAGE 83

83 Figure 5 5. Conceptual Diag ram of the Pathways through which Bolsa Familia Affects Child Mortality

PAGE 84

84 CHAPTER 6 DOES BOLSA FAMILIA REDUCE CHILD MORTALITY ? The purpose of Chapter 6 is to t est the hypothesis that, net of other independent variables that influence child survival, famili es who participate in the Bolsa Famlia program have lower rates of infant and child mortality compared to families that do not participate in the program. The first section describes the data that I will use, specifically the 2006 Pesquisa Nacional por Am ostra de Domiclios PNAD (the National Survey of Household Samples). The second section presents the design and methods of the analysis by addressing three main issues: it describes the depe ndent and independent variables, it summarizes the rationale for using logistic regress ion techniques, and it justifies the reasons for limiting the sample to a targeted sub population. The results of the analysis appear in the third section, followed by interpretations and conclusion. PNAD 2006 National health ins titutions typically use data collected in the official vital registration system in order to estimate infant and child mortality rates. To calculate the conventional rate, the denominator of the ratio is the total number of children born during a year. The numerator is the total number of children who died during the same period. The ratio is then multiplied by 1,000 in order to produce a standardized estimate. The conventional method of estimating infant and child mortality is fraught with a number of lim itations, especially in countries with a weak vital registration system. In Brazil, parents do not always register children who are born to them, nor do they always register the children who have died. The magnitude of the error is generally larger in the case of the numerator (deaths) compared to the denominator (births) of the ratio. Moreover, the relative magnitude of both errors varies geographically, such that the vital

PAGE 85

85 registration system is more accurate in the more developed regions and less accurat e in the less developed regions of the country. In addition to the questionable accuracy of official data on births and deaths, the vital registration system is also limited, at least for the purposes of analysis, by the fact that the records contain very little additional information beyond the date of the event and the age of child. In the face of these problems, demographer s estimate infant and child mortality on the basis of census and survey data. Unlike the vital registration system, which continuous ly collects information, censuses and surveys are conducted at a point in time. For this reason and for many others, censuses and surveys are subject to their own limitations. The main drawback stems from the fact that the exact date of an event a birth or a death is not known. This limitation, as it turns out, is not fatal, provided there is information on the mother sense, verified by sophisticated demographic techniques, confirms a close relationship between the age of mother and the ag e of her children: in the c ross section, younger women on average have yo unger children; older women on average have older children. If the age pattern of fertility is known, the exact age of children can be estimated with a high degree of precision. For the purposes of this study, the important point is that the PNAD household of live births and the number of children who are still alive at the time of the interview. Thi s information makes it possible to identify parous mothers who have experienced at least one child death (coded 1; 0 otherwise). Similarly, it also makes it possible to narrow the analysis to young women, 20 24 years of age, thereby ensuring that the risk of death is to younger rather than older children. With the individual level indicator of

PAGE 86

86 infant/child death in hand, I can model the outcome variable using the wealth of information contained in PNAD 2006.The Pesquisa Nacional por Amostra de Domiclios PNAD ( the National Household Survey is administered in Brazil every year by the Instituto Brasileiro de Geografia e Estatstica IBGE (the Brazilian Institute of Geography and Statistics), an agency of the federal government. The PNAD surveys collect da ta on more than 400,000 respondents including living conditions, educational attainment, occupational status, migration and family status. Because the sample size is so large, it means that many findings will be nationally representative, but also contain sufficient cases for carrying out sub national analyses. The 2006 PNAD dataset was appropriate for this study because, added to the core question, is a supplemental questionnaire which specifically asks respondents if they receive program benefits from Bol sa Famlia PNAD 2006 you or anyone in your household receive benefits from the Bolsa Famlia This information allows me to identify program participants. Other datasets ask respondents if they receive funds from a governme nt funded social welfare program but do not refer to the specific program. Also, PNAD contains additional socio economic and demographic information that allows me to control for variables that are known to influence child survival. Weighting the Sample To ensure that every region of Brazil is represented properly and proportionately population of Brazil. However, to maintain the original size of the sample (because expanding the number of cases to the entire population of Brazil would likely make all

PAGE 87

87 equivalent to the original weight multiplied by the sample interval (.0022053496). representative of the population yet is nonetheless the same size as the origina l unweighted sample. Logisitc Regression The dependent variable in the analysis that follows is a dichotomy. Women who have experienced a child death are coded 1; women who have not experienced a child death are coded 0. The objective is to use multivaria te techniques to model the probability of the outcome variable, in this case the probability of at least one child death. When the dependent variable is a dichotomy, ordinary least squares regression is not appropriate. Using a dichotomous dependent variab le in OLS regression violates the assumptions of normality and homoscedasticity (because a normal distribution is impossible with only two values). Also, when the values can only be 0 or 1, residuals (error) will be low for the portions of the regression l ine near Y=0 and Y=1, but high in the middle. Hence the error term will violate the assumption of homoscedasticity (equal variances). Even with large samples, standard errors and significance tests will be in error because of lack of homo scedasticity. Furt hermore for a dependent variable which assumes va lues of 0 and 1, the OLS model will allow estimates below 0 and above 1, which are outside the range of possible outcomes. Logistic regression is preferred because it enables the researcher to overcome many of the restrictive assumptions of OLS regression. Logistic regression does not assume a linear relationship between the dependents and the independents. The

PAGE 88

88 dependent variable need not be Logistic normally distributed. Normally distributed error terms are not assumed, and logistic regression does not require that the independents be interval. A logistic regression predicts the log of the odds of the dependent event. The natural log of the odds of an event is equal to the natural log of the probability of t he event occurring, divided by the probability of the event not occurring. Specifically: ln(odds(event)) = ln(prob(event)/prob(nonevent)) From a practical standpoint, logistic regression and least squares regression are almost identical inasmuch as both methods produce prediction equations, and in both cases the regression coefficients measure the predictive capability of the independent variables. However in logistic regression, the b coefficient is different. For every one unit increase in, say, the n umber of years of school completed, we expect an xxx decrease in the log odds of a child death. For dummy coded independent variables for example, when we code the presence of running water in the home 1 and its absence 0 the coefficient is the expecte d increase or decline (depending on the sign) in the probability of a child death associated with belonging to the effect category (having running water). In order to make the findi ngs easier to understand, the b coefficients are converted to odds ratios, by exponentiation (Exp (B)). Hypothesis Given the fact that: 1) there are specific requirements for program participation that relate to child health, 2) evidence suggests that the majority of families use the Bolsa Famlia money to purchase food and 3) previous studies find a inverse relationship between income and child mortality rates, I hypothesize that participating in the Bolsa Famlia program will be associated with a lower likelihood of having a child

PAGE 89

89 die. When analyzing whether the program has an impact on child mortality, it is necessary to try to control for other variables that could affect child mortality. Given the dataset that work through the proximate de terminants to influence the eff ects on child mortality. It is not possible, at least with the data at hand, to know the precise proximate mechanisms by which a particular socio demographic variable maternal education, for example influences child morta lity. The proximate determinant scheme thus remains framework nonetheless underscores the idea that the variables most often thought to affect the survival probabilities of children, such as income and education, do not have a direct effect on child survival but only operate through one or more of the proximate variables. The socio economic factors I chose are place of residence, region, age, ethnicity, education level (oper ationalized by literacy and years of school completed), socio economic status (operationalized by monthly per capita income) and living conditions (operationalized by access to running water in the house). My research hypothesis is that, controlling for pl ace of residence, region, age, ethnicity, literacy, years of school completed, per capita income and running water in the house, participants in the Bolsa Famlia program will be less likely to have a child die than non program participants. Therefore, the null hypothesis is that program participants will not be less likely to have a child die than non program participants.

PAGE 90

90 Dependent and Independent Variables For this study, I develop a binomial linear regression model to test the null hypothesis. The depe ndent variable is a calculated measure of having a child die. The rationale of creating this variable is detailed below. The main independent variable is program participation in Bolsa Famlia Program participation is a dummy variable whereby program part icipants are assigned a 1 and non program participants are assigned a 0. The aforementioned control variables include dummy variables and continuous variables. The dummy variables in the model are place of residence (where 0 means urban and 1 means rural), literacy where 0 means illiterate and 1 means literate, and running water where 0 means no running water in the household and 1 means running water in the household. Two of the independent variables, ethnicity and region, are set up to be dummy variables with reference categories. The variable used to measure ethnicity allows for 3 different categories, white, black and brown (the dataset also includes categories for yellow and indigenous but these individuals were excluded due to the small number of respo nses for those categories). I chose w hite Brazilian as the reference category to try to visibly demonstrate the disadvantage of being Afro Brazilian (either brown or black) in accordance with the literature on child mortality and race. I created two new va riables that measure the effect of being black or brown as opposed to being white. The final dummy variable I used is region which I created in much the same way as ethnicity. For the region variable, I use the Northeast as the reference category and the analysis measures the effect of being in any given region instead of being in the Northeast. For the dummy control variables, the effect found only shows the effect due to being part of the group assigned a 1 or not being part of that group (assigned a 0).

PAGE 91

91 The other control variables income and years of school completed are continuous variables. In a logi stic regression analysis, the b coefficient for continuous variables is an indicator of the effect of each increment in the independent variable on the dependent variable, in this case the probability of a child death. Both age and years of school are measured in years such that in the predictive model the effect found by the variable will be multiplied by the number of years. Monthly per capita income i s measured in reais (the Brazilian currency) and it is continuous and represented by an increase in 1 real for each increase of 1 in the data set. To calculate the dependent variable I first filtered the PNAD 2006 dataset to include only women with at lea st one live birth. This was done in order to limit the sample to mothers at risk of a chil d death, and to eliminate still births from the equation. Determining which women experienced the death of at least one of her children was not straightforward, given the nature of the PNAD 2006 questionnaire. Using the procedures described in Appendix A it was nonetheless possible to create a dichotomous (0 1) indicator, which could be used in a logistic regr ession framework. Results The frequency distribution o f the dependent variable shows t hat child mortality is not as common in Brazil as in the past. Of the women between the ages of 20 34 who gave live birth to a child, 389,112 have not had a child die. 19,441 women have had a child die. That means that in the PN AD 2006 dataset, 4.76% had a child die whereas 95.24% have not had a child die. Therefore, the sample does not have a high incidence rate of child mortality. Table 6 1 shows that households with running water are less likely to have a child die than house holds without running water. Of the households surveyed that have

PAGE 92

92 not had a child die, 10.65% do not have running water. 89.35% of the households surveyed that did not have a child die, do have running water. In contrast, of the households surveyed with a child that died, 16.13% do not have running water while 83.87% do have running water. This demonstrates the fact that having running water in the home is beneficial for the health of the family and reduces the likelihood of having a child die at a young ag e. The Chi Square value is 570.705 and the statistical significance is 0.000. This indicates that there is less than a 0.001% chance that the observed relationship is due to chance. Although it is general knowledge that having running water in a home is be tter for health outcomes, Table 6 1 shows the importance that the government should place on providing running water for all poor neighborhoods and homes in Brazil. For the analysis, I developed 5 different models with different conditions to determine if Bolsa Famlia has different effects for different groups of people in different regions of Brazil. The results of each model can be seen in Table 6 2. Model 1 limits the sample size to women between the ages of 20 34 who gave live birth to a child. I cho se this age range because demographers typically use this age range (or something very similar such as 18 3 4) for child mortality studies. Table 6 2 presents the findings of the likelihood of having a child die for all parous women in the sample aged 20 34. Many of the variables exhibit the expected relationship. For example, respondents from households with running water are less likely to have a child die at a young age. The Exp (B) is a value that shows the effect of the b coefficient on the dependent variable. Thus it should be multiplied by the likelihood of having a child die to determine the magnitude of the effect. If the Exp (B)

PAGE 93

93 value is equal to 1.0 then it has no effect. If it is less than 1, it has a reductive effect on the probability of havin g a child die. Conversely, if the Exp (B) value is greater than 1.0, the variable increases the probability of having a child die, net of the other variables in the equation. The Exp (B) value for homes with running water is 0.714 meaning that the likelih ood of having a child die would be reduced. Similarly, being literate and attaining a high level of education is associated with a reduced likelihood of having a child die. The Exp (B) value for literacy 0.698 is and the Exp (B) for level of educatio n is 0 .878. In contrast, being black or b rown is associated with a greater likelihood of having a child die. Black Brazilians are 1.393 and b rown Brazilians ar e 1.254 times more likely than w hite Brazilians to have a child die at a young age. Furthermore, being older also slightly increases the likelihood of having a child die at a young age. The age variable has an Exp (B) value of 1.069. The variable of interest does not exhibit the hypothesized association with likelihood to have a child die, however. Partici pants of the Bolsa Famlia program are more likely than non program participants to have a child die. The odd s ratio is 1.190, which indicates that, other things being equal, the probabilit y of a child death is 19 % higher among participants in the p rogram compared to non participants. Most of the variables are statistically significant at both the 0.10 and 0.05 levels of significance meaning that it is unlikely that these results were found due to chance. Interestingly, some are not statistically s ignificant, such as per capita income and some of the regional categories. That means that we cannot draw any conclusions about the effect of these variables on the likelihood of having a child die. Being literate and having

PAGE 94

94 access to running water in the home appear to be the 2 variables that have the strongest impact on reducing the probability of having a child die at a young age. Focus on Eligible Families The previous model gives us an idea of how all of the independent variables affect the likelihood of having a child die. However, the prior model conta ins data from wealthy women who do not qualify for the Bolsa Famlia program. To attempt to eliminate this bias, I made another model and filtered the sample to eliminate t he cases of any individuals wh o do not qualify for Bolsa Famlia For the purposes of this analysis, I define having a monthly per capita income of R$ 140 or less as constituting eligibility for program participation. For Model 2, the dataset was filtered to incl ude only women ages 20 34 who gave live birth and are eligible for Bolsa Famlia Some of the women are enrolled in the program and receive be nefits, whereas others do not. Note, I removed the monthly per capita income variable from the analysis in Model 2 since the entire sam The findings for Model 2 are very similar to Model 1. Again, homes with running water (Exp (B) = 0.751 ), and respondents who are literate (Exp (B) = 0.697) and have obtained a higher level of education (Exp (B) = 0.886) are less likely to have a child die. Moreover, being older and being black or brown is associated with a higher probability of having a child die ( Exp (B) values of 1.088, 1.299 and 1.296 respectively). Finally, the program par ticipation variable for Bolsa Famlia has an Exp (B) value of 1.015, but it is not statistically significant. Focus on Younger Women Models 1 and 2 may not capture the effects of Bolsa Famlia due to the nature of the sample. In the PNAD 2006 data we ca n know whether a woman had a child die and

PAGE 95

95 whether she receives benefits from Bolsa Famlia but we cannot know if the child died before the family began participating in Bolsa Famlia The timing problem is larger among older women if only because, on aver age, their children are older. As a consequence there is a greater chance that a woman who reports a child death may be referring to an event that took place 5 or 10 years ago. In an attempt to make sure that the beginning of participation in Bolsa Famli a and the experience of mortality are as close as possible, I restricted the sample to women aged 20 24 who are eligible for Bolsa Famlia for Model 3. That way I am more likely to capture children who were born into a family already participating in Bols a Famlia or at least have spent the majority of their first 5 years in the program. Therefore, the intent is to examine if the program has a stronger or a weaker effect on the likelihood of having a child die for a more restricted age group of women. Fro m Model 3, it can be observed that the Exp (B) value for the Bolsa Famlia variable is now 0.916, but the result is not statistically significant. However, the fact that t he Exp (B) value in Model 3 is reduced suggests a possible association between progra m participation and a lower likelihood of having a child die. The other variables demonstrate the expected association with the probability of having a child die at a young age and some are statistically significant. Strikingly, neither one of the categori es for ethnicity (b lack in comparison to white and brown in comparison to w hite) is statistically significant. Focus on Regions Based on the history of child mortality in Brazil ( C hapter 5 contains further details), the country can be separated into 2 regi ons that vary in terms of child mortality rates: a Northern region, which would consist of the North and the Northeast, and a

PAGE 96

96 Southern region which consists of the Southeast, the South and the Central West regions. Figure 6 1 visually represents these two conceptual mortality regions of Brazil. Traditionally, the Northern region has had higher child mortality rates than the Southern region. Models 1, 2 and 3 all examined the likelihood of having a child die for the entire country of Brazil while controlling for region. Models 4 and 5 however carry out the analysis within There are two competing hypotheses that justify the analysis based on regional diff erences that are potentially relevant to the Bolsa Famlia effect. The first is based on the observation that, small increases of income lead to large declines in mortality in high mortality populations. Conversely, in low mortality populations, small incr eases in i ncome do not have much effect. This is due to the predominant causes of death differ in high and low mortality groups. In high mortality populations, a large proportion of children die of diseases and infections such as dysentery, pulmonary disor ders, diarrhea and a variety of othe Exogenous causes of death are relatively easy to control given technology and medical advances. Hence, small increases in living standards lead to sharp declines in mortality. Conversely when mortality is low the main causes of death tend to be factors because the majority of deaths due to exogenous factors have already been prevented to reach the status of a endogenous factors are difficult to control and therefore reducing the mortality rate any

PAGE 97

97 further is unlikely. Under these circumstances, increases in income do not have much effect. Differences in th e causes of death in high and low mortality populations are relevant to this study because they suggest reasons for regional differences. Specifically, the exogenous/endogenous distinctio n leads me to anticipate that Bolsa Famlia may be more likely to ha ve an effect in t he North, where mortality is high and sensitive to economic status, and may be less likely to have an effect in th e South, where mortality is low and less sensitive to economic status. A competing hypothesis predicts the opposite that B olsa Famlia would have a larger impact in the more developed regions of the country. This line of reasoning focuses, not on the structure of cause of death, but on the institutional differences in the quality of health care services. The healthcare infras tructure and access to technology is more widely available in the Southern region of Brazil. Furthermore, the quality of services is an important consideration because one of the main channels through which Bolsa Famlia can influence child survival is thr ough the specified he alth related conditionalities. Women who are Bolsa Famlia recipients are required to have pre natal examinations. Children are required to be immunized and visit a health clinic for check ups. But the effectiveness of these activities in terms of reducing mortality may be contingent on the quality of the health services available, which are much better in the more developed regions of the country. Put another way, Bolsa Famlia recipients in the Northern and in the Southern regions may both adhere to program requirements, yet the benefit of adhering to the health related requirements would be greater in the more developed regions of the country. Given this logic, it is conceivable that Bolsa Famlia

PAGE 98

98 would have a more positive impact on child mortality in the Southern region than in the Northern region. Table 6 3 shows the same logistic regression for two additional models that separate respondents by region. The sample for Model 4 is limited to women ages 20 24 who are eligible for Bolsa Famlia and live in the High Mortality region (the North and the Northeast). The Exp (B) value for Bolsa Famlia is 1.101 which indicates that program participants are slightly more likely to have a child die at a young age, but the results are not statistically significant. Surprisingly in Model 4, many variables do not display the expected relationship. For example, for the different ethnic categories, brown and black Brazilians (with Exp (B) values of 0.682 and 0.888 respectively) are slightly le ss likely to have a child die at a young age than white Brazilians (the reference category ) Yet these findings are not statistically significant. Only two variables are statistically significant, age (Exp (B) value of 1.198) and literacy (Exp (B) value of 0.303) and they continue to represent the same thing they have in previous models. An increase in the age of the mother is associated with an increase in the likelihood of having a child die at a young age while being literate is associated with a decrea sed likelihood of having a child die. Based on the results of Model 4, we cannot accept the hypothesis that Bolsa Famlia is most likely to have a reductive effect on child mortality rates in high mortality regions. Model 5 in turn provides the most positi ve results of any of the models. The sample for Model 5 is limited to women ages 20 24 who are eligible for Bolsa Famlia and live in the Low Mortality region (the Southeast, South and Central West regions). For the first time, the model finds that Bolsa Famlia has a statistically significant

PAGE 99

99 mortality reducing effect. For this model, the Bolsa Famlia variable has an Exp (B) value of 0.244 meaning that in the South region Bosla Famlia reduces the probability of a child dea th by around 75 % (1 .244= .756). How then can we best interpret the results of Model 5? Remember that because loped healthcare infrastructure; it stands to reason that this infrastructure plays a significant role in determi nin g the child mortality rate. PNAD data are not capable of measuring what is responsible for the association between Bosla Famlia and the child mortality rate, but we can conclude that the health conditions of program participation might make a difference f or many program participants. Perhaps it is the fact that pregnant women are receiving better pre natal care and are thus giving birth to healthier children. Or perhaps the money from the cash transfer is being used for better food which in turn augments t nutritional status. Either way, it is clear that Bolsa Famlia is having a positive impact on the lives of young children in the low mortality, Southern region of Brazil. In accordance with the research hypothesis, net of the effects of place of residence, age, race, educational attainment, literacy and access to running water, in Southern Brazil Bolsa Famlia participants are less likely to have a child die than non program participants. Interpretation The results of Model 5 do support the resea rch hypothesis. However, similar to much of the literature examined about the effects of program participation on health outcomes, it would appear that Bolsa Famlia is not 100% successful in having the intended impact on child mortality rates in Brazil. T he findings that Bolsa Famlia has a that healthcare infrastructure is important and that program conditions cannot be

PAGE 100

100 properly met without a well developed infrastruc ture. For the underdeveloped North and Northeast regions of Brazil, building a better healthcare infrastructure would require a large amount of money. It would be premature to conclude that investment in healthcare infrastructure is the answer to the probl Northern region. In regards to the negative findings for M odels 1 4 bear in mind that Bolsa Famlia was not designed as a program to reduce child mortality rates. Many different factors contribute to the chi ld mortality rate of a country. These factors include access to running water, access to health clinics and medical services, and nutritious foods. All of these factors are controlled outside of the Bolsa Famlia program. The Brazilian government has many other public health programs that are more specifically targeted to tackling health problems. Furthermore, these findings could suggest that Bolsa Famlia is properly targeted to serve the most vulnerable population of Brazil. If families are poor and more likely to have a child die because they are living in poverty and cannot provide nutritious food or proper healthcare for their children, these are the families that should be enrolled in Bolsa Famla and receiving the benefits. This justification does no t quite explain the slight advantage of non program participants at the same per capita income level of program participants found in M odels 2, 3 and 4. However, the data do not tell us if some respondents qualify to participate in the program yet still ha ve a greater per capita income than the average Bolsa Famlia participant. We cannot know for sure if there is a source of bias that we are missing.

PAGE 101

101 To improve the models further, I could continue adding explanatory variables but that would only further co mplicate the model without adding much strength to it. The explanatory variables already include d do a good job of covering socio economic and demographic indicators that act through proximate determinants to impact child mortality. Either way, it is clear that there is still room for improvement in the Bolsa Famlia program.

PAGE 102

102 Table 6 1. Having a child die by running water in the h ome Brazil 2006 Have you had a child die? Running Water in the Home No Yes No 41,577 (10.65%) 3,136 (16.13%) Yes 348,68 6 (89.35%) 16,310 (83.87%) Total 390,263 (100.00%) 19,446 (100.00%) Chi square = 570.705, ss = 0.000 Source: PNAD 2006

PAGE 103

103 Table 6 2. Odds ratio of having a child die by p articipation in Bolsa Famlia, age, ethnicity, educational attainment, literacy, pe r capita income, running water, place of residence and r egion: Brazil 2006 Model 1 : Women ages 20 34 Model 2 women ages 20 34 Model 3 women ages 20 24 b coefficient Exp (B) b coefficient Exp (B) b coefficient Exp (B) Bol sa Famlia 0.174* 1.190 0.014 1.015 0.088 0.916 Age 0.067** 1.069 0.084** 1.088 0.218** 1.243 Ethnicity White (ref) Black 0.332** 1.393 0.262 1.299 0.129 1.138 Brown 0.226** 1.254 0.259* 1.296 0.038 1.038 Years of School 0 .130** 0.878 0.121** 0.886 0.103* 0.902 Literacy 0.359** 0.698 0.361** 0.697 0.961** 0.383 Per Capita Income 0.000 1.000 Access to Water 0.336** 0.714 0.286* 0.751 0.337 0.714 Place of Residence Urban (ref) Ru ral 0.008 1.008 0.009 0.991 0.110 1.116 Region Northeast (ref) North 0.115 1.122 0.092 1.097 0.075 0.928 Southeast 0.450** 0.638 0.379** 0.684 0.778* 0.459 South 0.187 0.829 0.160 0.852 0.099 1.104 Central West 0.420 ** 0.657 0.381* 0.683 0.631 0.532 Source: PNAD 2006 Notes: Model 1: R 2 = .094 ; Model 2: R 2 = .087 ; Model 3: R 2 = .107 Statistical significance: = 0.05, ** = 0.001 $R 140

PAGE 104

104 Table 6 3. Odds r atio of having a child die by p articipation in Bolsa Famlia, age, ethnicity, educational a t tainment, literacy, running water and place of r esidence: Brazil 2006 Model 4 : High Mortality ages 20 24 Model 5 : Low Mortality women ages 20 24 b coefficient Exp (B) b coefficient Exp (B) Bolsa Famlia 0.096 1.101 1.409* 0.244 Age 0.180* 1.198 0.461* 1.585 Ethnicity White (ref) Black 0.383 0.682 1.226 3.407 Brown 0.118 0.888 0.240 1.272 Years of S chool 0.056 0.945 0.307** 0.736 Literacy 1.193** 0.303 0.195 0.822 Access to Water 0.198 0.820 1.565* 0.209 Place of Residence Urban (ref) Rural 0.175 1.192 0.066 1.069 Source: PNAD 2006 Notes: Model 4: R 2 = .082 ; Model 5: R 2 = .2 18 Statistical significance: = 0.05, ** = 0.001 $R 140

PAGE 105

105 Figure 6 1. Map of B razil with c onc eptual c omparative r egions of h igh c hild m ortality and l ow c hild m ortality

PAGE 106

106 CHAPTER 7 CONCLUSIONS AND FUTURE WORK As this study demonstrates, Bolsa Famlia is a unique poverty alleviation program that provides the incentive for participants to develop their own human capital while providing them with cash stipends to give them some additional income. Bec ause of its required health conditions and the demonstrated correlation between income and child mortality rates, it was hypothesized that net the effects of other factors including age, race, level of education, etc. participating in Bolsa Famlia would reduce the likelihood of having a child die at a young age. The results, based on logistic regression analyses, showed that the anticipated mortality reducing effect was observed only among eligible women, 20 to 24 years of age, living in the South ern, low mortality region. The same reductive effect was not found for the Northern, high mortality region or for Brazil on a national level. However, this does not necessarily mean that the Bolsa Famlia program is a failure. The problem might not be in the progr am itself, but rather in the data used for the analysis and the inability of the data to truly capture the encouraging and suggests that Bolsa Famlia can still b e improved to have a similar effect on child mortality in all regions of Brazil. One of the main differences between the Northern, high mortality region and the Southern low mortality region is that the healthcare infrastructure is more developed in the So largest and wealthiest cities. Therefore, it follows to reason that the hospitals and health clinics in the cities of the Southern region would have the best access to technology and

PAGE 107

107 medicines. Furthermore, it is also conceivable that the states in the Southern region have invested more in healthcare infrastructure tha n states in the Northern region; however we cannot be certain without researching the healthcare infrastructure i n Brazil. A n analysis of the Brazilian healthcare infrastructure would examine several indicators including but not limited to number of hospital beds per 1,000 people, number of doctors per 1,000 people, amount of money invested in healthcare per capita etc. On a regional basis such an analysis could reveal if the healthcare infrastructure in the Southern region is truly more developed than the Northern region Such a study would be very intensive, but could help lend support to the hypothesis that Bols a Famlia seems to be more effective at reducing child mortality in the Southern region due to a better developed healthcare infrastructure. Remember that the leading causes of infant death in Brazil are problems originating during the perinatal period. So me of these problems occur due to improper or lack of pre natal care. One suggestion for improving the Bolsa Famlia program and its effects on child mortality, is to insure that pregnant women receive proper pre natal care. As the program currently stands pregnant women could be going to receive pre natal care, but are not receiving adequate attention We have no way of knowing if each hospital and health clinic in Brazil is adequately equipped with the staff and the tools that it needs to properly care f or pregnant women. However, devoting more attention to pre natal care could lead to a reduction of child mortality due to perinatal causes. An overview of the literature reveals that many authors could not find the correlation they were hoping for between Bolsa Famlia and positive health outcomes. Does this indicate that there is a flaw in the Bolsa Famlia program design? Not

PAGE 108

108 necessarily. Bolsa Famlia is but one of many different programs from the Brazilian federal government. Many other programs specif ically target health outcomes and might be more suited to bring about the desired changes in child healthcare in Brazil. Since the 1930s, Brazil has made progress in reducing the infant mortality rate and it continues to do so, approaching the same levels observed in many developed countries. However, regional disparities remain. The Northeast region has the highest infant mortality rate compared to the South, which has the lowest. Regional inequality in Brazil extends far beyond health inequalities and als o includes income inequality. There are many areas for future research concerning Bolsa Famlia and conditional cash t ransfer Programs in general. Due to the el ection of Dilma Rousseff, will continue to be in power in Brazil and Bolsa F amlia will continue overty alleviation strategy. Two needed area s of research are the political implications of Bolsa Famlia and its possible use as a clientelistic policy. Is Bosla Famlia a form of a political patron c lient system? If so, what does future expansion of the program mean for the PT and politicians in Brazil? Another area of research is the coverage of the program. Bolsa Famlia appears to be fairly well targeted, yet does not includ e all people who are el igible. Some studies suggest that program expansion would effectively cover all poor families. But what are the consequences of expanding Bolsa Famlia and would it be worth the price to greatly increase program coverage? Finally, how can Bolsa Famlia be better adapted to meet its goals with regard to health and educational outcomes? Specifically, is there any change that can be made to program requirements or implementation to increase compliance with program requirements? What can be learned from other c onditional

PAGE 109

109 cash t ransfer programs to help adapt Bolsa Famlia to achieve specific health goals? All of these questions, and many more, still need to be answered to fully understand the implications and the possibilities of the Bolsa Famlia program

PAGE 110

110 APPEND IX EXPLANATION OF CREATING A VARIABLE IN PNAD 2 006 TO MEASURE THE EXPERIENCE OF CHILD DEATH One of the fundamental variables for the analysis is whether or not a woman has had a child die. This variable can be used to determine if participants in the Bolsa Famlia program are more or less likely than program non participants to have a child die. However, PNAD does not measure whether or not a woman had a child die directly. Instea d it can be calculated by using the responses from some of the other questions in the survey. the responses to these questions, I was able to first calculate the number of children living inside the home by summing the number of male and female children living insid e the home for each female respondent. In so doing, I created a new variable labeled calculated the number of children living outside the home by adding the number of mal e children to the number of female children living outside of the home. I created a variable that live outside of the home. Note that this is not to imply that these chil dren are homeless. It is possible that they live with another relative or that there is another explanation for the fact that they live outside of the home. Regardless, the dataset does not afford us enough information to know why these children live outsi de of the home.

PAGE 111

111 I followed the same process to calculate the number of children that the respondent ha d given a live birth to but who had already died. PNAD includes questions which represents the total number of live children that the woman had who children die a 0. It also assigned every woman who gave l ive birth and had at least 1 child die a value of 1 (even if the woman had more than 1 child die). Thus, when a live who are still living and a 1 represents that at least 1 c hild born a live to the respondent has died.

PAGE 112

112 LIST OF REFERENCES Northwestern University, Evanston, IL. Atta nasio, Orazio, Lus Carlos Gmez, Patricia Heredia, and Marcos Vera Hernndez. Term Impact of a Conditional Cash Subsidy on Child Health Studies, London. Barham Tania. 2005. Providing a Healthier Start to Life: The Impac t of Conditional Cash Transfers on Infant Mortality University of California at Berkeley, Department of Agriculture and Resource Economics. Barros, Ricardo Paes de, Mirela de Carvalho and Rosan o 1414 Instituto de Pesquisa Econmica Aplicada. Rio de Janeiro: Instituto de Pesquisa Econmica Aplicada. Bautista, Sergio, et al. 2004. bilidad y el Estado de Salud de la Poblacin Beneficiaria y en la Utilizacin de Servicios de Salud: Resultados de Corto Plazo en Zonas Urbanas y de Mediano Plazo en Zonas Evaluacin Externa del Impacto del Programa de Desarrollo Humano Oportunid ades 2004: Salud edited by Bernardo Hernndez Prado and Mauricio Hernndez vila, vol. 2, chap. 1. Cuernavaca, Mexico: Instituto Nacional de Salud Pblica. Behrman, Jere R., Susan W. Parker, and Petra E. Todd. 2009. Term Impacts of the Oportunida des Conditional Cash Transfer Program on Rural Youth in 270 in Poverty, Inequality, and Policy in Latin America edited by Stephan Klasen and Felicitas Nowak Lehmann. Cambridge, MA: The MIT Press. sila, DF: Ministrio de Desenvolvimento Social e Combate Fome, Retreived Nov. 17, 2010 ( http://www.mds.gov.br/bolsafamilia ). to Social e Combate Fome, Retreived Nov. 17, 2010 ( http://www.mds.gov.br/bolsafamilia/beneficios ). Social e Combate Fome, Retreived Nov. 17, 2010 ( http://www.mds.gov.br/bolsafamilia/condicionalidades ).

PAGE 113

113 Socia l e Combate Fome, Retreived Nov. 17, 2010 ( http://www.mds.gov.br/bolsafamilia/cadastrounico ). Retreived Jan. 16, 2011 ( http://www1.folha.uol.com.br/folha/dimenstein/colunas/gd030706.htm ). Fiszbein, A., Schady, N., Ferreira, F., Grosh, M., Kelleher, N., Olinto, P., et al. 2009. Co nditional Cash Transfers: Reducing Present and Future Poverty Washington, D.C.: The World Bank. Franko, Patrice. 2007. The Puzzle of Latin American Economic Development Lanham, Maryland: Rowman & Littlefield Publishers, Inc. Gertler, Paul. 2000. Final Report: The Impact of Progresa on Health Washington, D.C.: International Food Policy Research Institute. Gertler, Paul. 2004. Do Conditional Cash Transfers Improve Child Health? Evidence American Economic Re view 94 (2): 336 41. Gomes Victoria, Cesar, and Fernando Celso Barros. perinatal causes in Brazil: So Paulo Medical Journal 119, no. 1: 33 42. Hernndez, Daniel, and M aria del Carmen Huerta. 2000. Reproduct Evaluacin de Resultados del Programa de Educacin, Salud, y Alimentacin : 43 80. Washington, D.C.: International Food Policy Research Institut e. Rio de Janeiro: Instituto Brasileiro de Geografia e Estatstica, Retreived December 7, 2010 ( http://www.ibge.gov.br/english/presidencia/noticias/noticia_visualiza.php?id_noti cia=1293&id_pagina=1 ). Levy, Santiago. 2006. Oportunidades Program Washington, D.C.: Brookings Institution Press. Fome, Retreived Sept. 12, 2010 ( http://www.mds.gov.br/saladeimprensa/noticias/2010/setembro/846 mil beneficiarios do bolsa familia precisam atualizar dados ate fim de outubr o ).

PAGE 114

114 Medeiros, Marcelo, Tatiana Britto, Fbio Soares. Discusso N o 1283 Instituto de Pesquisa Econmica Aplicada. Braslia : Instituto de Pesquisa Econmica Aplicada. Molyneux, Maxine. Social Policy and Administration 40, no. 4: 425 449. Morris, Saul, Pedro Olinto, Rafael Flor es, Eduardo A.F. Nilson, and Ana C. Figueir. Journal of Nutrition 134: 2336 41. e of Premarket Factors in Black White Wage Journal of Political Economy 104, no. 5: 869 895. 22 in New Left Review 42, edited by Susan Watkins. London. PNAD (Pesquisa Nacional p Brasileiro de Geografia e Estatstica. Rocha, Sonia. 2008. O Brasil Dividido: Espacializao Alternativa e Pobreza Rio de Janeiro, Brazil: Publit Solues Editoriais. York Times, Retrieved Jan. 25 th 2011 ( http://opinionator.blogs.nytimes.com/2011/01/03/to beat back poverty pay the poor/ ). American Journal of Public Health 64, no. 8: 792 798. Signorini, Bruna Atayde and Bernardo Lanza Queiroz. Program in the Benefic Working Paper. Belo Horizonte: UFMG CEDEPLAR. Socioeconmica. Instituto Brasileiro de Geograf ia e Estatstica: Rio de Janeiro. Skoufias, Emmanuel. 2000. Is Progresa Working? Summary of the Results of an Evaluation by IFPRI. Washington, D.C.: International Food Policy Research Institute.

PAGE 115

115 Soares, Sergei, Rafael Guerreiro Osrio, Fbio Veras Soares Marcelo Medeiros, Discusso N o 1293 Instituto de Pesquisa Econmica Aplicada. Braslia: Instituto de Pesquisa Econmica Aplicada. Cobertura do Programa Bolsa Famlia: Qual o Significado dos 11 Milhes de o 1396 Instituto de Pesquisa Econmica Aplicada. Rio de Janeiro: Instituto de Pesquisa Econmica Aplicada. o 1424 Instituto de Pesquisa Econmica Aplica da. Braslia: Instituto de Pesquisa Econmica Aplicada. Dissertao. Belo Horizonte: UFMG CEDEPLAR. United Nations Development Programme. 2009. Human Development Report 2009 Retreived: Nov. 15, 2009 ( http://hdrstats.undp.org/en/indicators/161.html ). Vaitsman, Jeni and Paes Sousa, Rmulo. 2007. Avaliao de Polticas e Programas do MDS Resultados: Volume 2 Bolsa Famlia e Assistncia Social Brasilia, D.F.: Secretaria de Avaliao e Gesto da Informao, Ministrio do Desenvolvimento Social e Combate Fome. Weissheimer, Marco Aurlio. 2006. Bolsa Famlia: Avano s, Limites e Possibilidades do Programa que Est Transformando a Vida de Milhes de Famlias no Brasil So Paulo: Editora Fundao Perseu Abramo. Wood, Charles H. and Magno de Carvalho, Jos Alberto. 1988. The Demography of Inequality in Brazil Cambridg e: Cambridge University Press. The World Bank Online. 2010. WDI Online: World Development Indicators Washington, D.C.: The World Bank, Retrieved October 25, 2010. ( http://data.worldbank.org/country/ brazil ).

PAGE 116

116 BIOGRAPHICAL SKETCH Clay Giese was born in 1987 in Morganton, North Carolina. The older of two children, he grew up mainly in Valdese, North Carolina, graduating from East Burke High School in 2005. He completed his BA in Latin A merican Studies and Spanish at Vanderbilt University in May 2009. Furthermore, Clay earned is MA in Latin American Studies from the University of Florida in May 2011. His concentration within Latin American Studies is Development Studies. Clay has an inter est in poverty alleviation programs throughout the region and the outcomes of conditional cash t ransfer programs.