Moving Out of Poverty: Growth and Freedom from the Bottom-Up

Transcription

1 Moving Out of Poverty: Growth and Freedom from the Bottom-Up Research Proposal Global Development Network Governing Body January 27, 2004 By Deepa Narayan, Senior Advisor, PREM World Bank INTRODUCTION... 2 OBJECTIVES AND SCOPE... 4 ANALYTIC FRAMEWORK AND HYPOTHESES... 5 SAMPLING DATA COLLECTION INSTRUMENTS DATA ANALYSIS WORK PROGRAM OUTREACH PROPOSED THREE-YEAR BUDGET ANNEX LITERATURE REVIEW, SUMMARY TABLES AND BIBLIOGRAPHY LITERATURE REVIEW Mobility Panel Studies Mobility Other Quantitative and Qualitative Studies Social Groups and Social Identity Democracy, Good Governance and Poverty Reduction

2 Introduction Study Purpose: To conduct a global comparative study in 20 countries to learn retrospectively from those who were once poor but have moved out of poverty and stayed out of poverty and those who have remained poor in different political, social and economic environments. Why and how do some poor people move out of poverty and stay out of poverty while others remain trapped in chronic poverty? Moving out of Poverty is a global 20- country study to explore the dynamics of five-to-ten year transitions out of poverty and chronic poverty from the perspectives of the men, women and youth who have lived these experiences. Fundamentally, the study seeks to enhance understanding of factors that unleash poor people s economic potential and support their transitions out of poverty. The global breadth of the study will enable examination of economic, social, political and institutional factors across different environments. There is broad agreement that growth that benefits the poor is critical for poverty reduction on a large scale. But there is relatively little agreement on how to achieve it in countries with very different initial economic, political and social conditions. Most of our knowledge is based on household surveys that focus primarily on household economic and demographic characteristics, rather than on social, political and institutional conditions that also shape access to opportunities. A review of the economic literature based on household surveys, cross-country regressions, and panel studies confirms that the cluster of factors most often associated with growth that benefits the poor are: household endowments; geography; economywide factors including macroeconomic volatility and trade policies; and initial levels of inequality in assets, land, and education and literacy. Recent studies using panel data and poverty maps over time contribute further to understanding sources of growth and the 2

3 role of inequality in initial conditions. However, researchers state over and over again that measurement error is likely to be high in these studies. Further, these studies do not examine interaction among variables, nor do they usually include social, institutional or political factors. The number of democracies around the world has doubled over the last couple of decades. Overall, democracies have been slow and steady but not spectacular in reducing poverty. Hence, despite normative assumptions about democracy, it is unclear whether mobility strategies out of poverty and the experience of economic freedom and choice vary significantly across political and governance systems. How do democracy and governance systems impact poor people's experiences of economic freedom and choice at the local level? This study seeks to address these questions and provide policy insights based on the experiences of those who have actually moved out of poverty and stayed out of poverty over time, and those who have stayed trapped in chronic poverty. The focus on the wide range of social, political, institutional and economic mechanisms that hinder or facilitate poor people s access to economic opportunity and movement out of poverty will complement ongoing work at the more macro level. There is strong relevance of this study to the design and monitoring of poverty reduction strategies and policies in developing countries. Policymakers and local people s perspectives of key events and policies that have impacted poverty reduction over the last decade will be compared. The study will also provide insights on people s own definitions and understandings of mobility, freedom, power and aspirations, and how these concepts are linked to poverty reduction. 3

4 The study seeks to advance mixed-method research approaches by maximizing use of existing panel and cross sectional studies or poverty maps, and by ensuring strong integration of the analytic and sampling frameworks for both qualitative and quantitative data collection. The research program will engage diverse actors and seek to support local learning and action and links to national policy arenas, including PRSP processes where relevant. Objectives and Scope The study is primarily concerned with examining the dynamics of five- to ten-year transitions out of poverty or staying in chronic poverty from the perspectives of those who have lived through these experiences. The global breadth of the 20-country study will enable examination of mobility and chronic poverty over time across different political, social, and economic environments. The study will take place in countries with and without rapidly growing economies, with strong and weak democracy and governance systems, and with different degrees of openness to trade and globalization. The key questions that the study will examine are: How and why do some men and women move out of poverty and stay out of poverty while others remain trapped? Do people experience mobility differently under contexts of fast and slow economic growth? Does the extent of global integration affect people's mobility? Are there gender differences? Do networks and social identity matter in men's and women's upward climbs? Do the quality of local governance, "depth" of democracy and freedom make a difference? In trying to answer these questions, the study will explore people s understandings and definitions of mobility, freedom, power and aspirations-- and the importance of these concepts in their lives. 4

5 Given that poverty and well-being are multidimensional, the study seeks to uncover the dimensions which are considered to be most critical to moving out of poverty or being stuck in chronic poverty. Interactions among the various dimensions and patterns in the sequencing will be examined. In addition, individual life histories will be related to key events and policy shifts at the community and national level. The study will explore gender differences in the triggers for escaping poverty and in the dynamics of chronic poverty. Depending on available data sets at the country level, the study will explore movements over a five to ten year period, rather than shorter-term fluctuations in income and wellbeing. The focus will be principally on the upwardly mobile, compared with the chronic poor. Given that much is already known about the triggers that push households down into poverty, only a small sample may be included of those falling into poverty. Analytic Framework and Hypotheses The basic proposition of the study is that transitions out of poverty are determined by a broad set of economic, social and political factors. To turn this into testable hypotheses, a number of specific potential determinants of transitions have been identified within each of the three types of factors. It is also recognized that these factors interact with each other. These hypotheses are based on workshop discussions, recommendations and a literature review (see Annex). 5

6 Economic Factors. Whether a specific household is able to escape from poverty is a function of the macroeconomic environment in which it lives, the economic characteristics of its community, and the household's asset endowment and outlook on the future. While this study will not formally test the role of macroeconomic variables (for example, by means of cross-country regressions), the importance of these variables is recognized by virtue of the fact that three of the four variables used to stratify the selection of countries reflect macroeconomic conditions: level of income, rate of growth and share of exports in GDP. This will make it possible to determine whether the nature of the poverty transitions and their key determinants are influenced by these macroeconomic conditions. A similar argument applies to the economic conditions of the community. We will examine the patterns in the transitions and their determinants according to the economic performance of the community. Specific attention will be given to communal infrastructure (schools, health facilities, roads, electricity, communications, etc.) to assess whether households living in better endowed communities find it easier to escape poverty on a permanent basis. At the household level, the hypothesis is that the household's ability to escape poverty is a function of the asset endowment of the household, the returns to these assets, and the household's ability to manage over-time risk of income fluctuations. The economic literature on poverty and risk management has well documented the importance of these three factors. 1 The household's ability to escape poverty on a permanent basis is viewed as being determined by the same three factors, although the relative importance of each factor is likely to be different. Specifically, the ability to 1 See, for example, World Bank

7 avoid catastrophic income downturns in response to exogenous shocks plays a key role in keeping households out of poverty. The household survey that is part of the study will collect information on each of the factors. Asset information will cover physical assets (productive assets such as land, cattle, tools and equipment, etc.; housing, housing amenities and household durables) and human capital. The returns to these assets will not be captured directly in the survey because of the high degree of difficulty in collecting reliable income information. However, it is known that these returns are determined by the functioning of markets and by the household's ability to access these markets. Thus, the survey will collect detailed information on the household's market participation and the constraints it faces in doing so (for example, discrimination, lack of information or absence of markets). The household will also be asked about different sources of income (farm and non-farm income, wage income, household enterprise income, transfer income) because income diversification is an important tool of risk management. Likewise, detailed information will be asked about access to credit and the household's situation of indebtedness. The open-ended interviews will explore in-depth the mechanisms that have helped people move out of poverty or kept them trapped in chronic poverty. In summary, the basic economic hypothesis that will be tested by the study is that households have a higher probability to escape poverty on a permanent basis if they (a) live in communities with better infrastructure endowments and better overall economic performance, and (b) have higher asset endowments, better access to markets and credit, and more diversified income sources. 7

8 Social Factors. The study will probe into the importance of social assets (social capital) and social structure. At the community level, social capital is manifested by the existence of strong community organizations that are inclusive and well-connected to the outside world, and by the existence of norms of trust and solidarity. Such social capital facilitates information sharing and mutually beneficial collective action and decision making. 2 This has the potential to enhance the returns of household assets and to provide a measure of collective risk management to complement risk management at the household level. High levels of social cohesion and trust also help communities to deter crime and violence. In addition, social capital has the potential to reduce gender differences and the barriers to making opportunities available to all social groups in the community. The hypothesis is thus that community social capital facilitates household efforts to escape from poverty, particularly if the members of these networks span different social and economic groups. In addition, there is growing empirical evidence that, at the household level, social capital is also associated with higher welfare, reduced poverty, and better asset accumulation and risk management. 3 While these studies provide strong evidence of a positive correlation between social capital and various outcome variables at the household level, the evidence on the direction of causality is weaker because the studies are all based on cross-sectional data. By the same token, panel-based studies of poverty transitions have typically not had available data on social capital. 4 This study will thus attempt to test formally the hypothesis that households with more social capital have a greater probability to escape poverty on a permanent basis. This hypothesis complements 2 See, for example, Narayan 1995, Hino 1993, Mansuri and Rao See, for example, Narayan and Pritchett 1997, Grootaert 2001, Grootaert and van Bastelaer A notable exception is Maluccio et al

9 the similar hypothesis on the role of social capital at the community level. The household survey that is part of the study will collect detailed information on different dimensions of social capital (membership in local organizations, trust and solidarity, collective action, etc.). Evidence on trends in these variables will be collected both in the survey and as part of the qualitative community information. In summary, the basic social hypothesis that will be tested by the study is that households have a higher probability to escape poverty on a permanent basis if they (a) live in communities with stronger community organizations and norms, and (b) have higher levels of social capital. Political Factors. A country's political system affects households and their ability to escape poverty in multiple direct and indirect ways. 5 This study focuses on four intermediate variables: democracy, governance, power and freedom and aspirations. At the macro level, the study will focus on the role of governance and democracy, to test the hypothesis that poverty transitions are facilitated in countries with well-functioning democracies and good governance. While the study will not be able to provide a formal test whether democracy is the preferred system for poverty reduction, it is expected to document differences in the nature of poverty transitions, especially in prevailing bottlenecks, between democratic and undemocratic countries. In general though, it is expected that the importance of democracy and governance will be most evident at the level of the community and local government. In this context, it is useful to unpackage the two concepts. Although both democracy and governance are concerned with the selection, oversight and replacement of public leaders, 5 See, for example, Diamond

10 these aspects will relate more directly to understanding the forms and quality of local democratic functioning and political inclusion. The governance aspects of the study will focus on the extent to which public resources serve the public good rather than particularistic interests. The study will also, for instance, be exploring the dynamics of economic and political processes in countries that may have closed political systems at the national and/or local levels, but have nevertheless attained large poverty declines (which can perhaps be traced to strong public sector capacities at the central or decentralized level, in addition to other factors). At the community level, it is hypothesized that democratic elections and other participatory forms of selecting and influencing public leaders promotes poverty reduction because these channels provide valuable openings for poor people and their networks to advance their interests and hold authorities accountable. Good governance promotes escaping from poverty because it eliminates the cost of corruption (which is especially burdensome for the poor), improves the delivery of public services, protects law and order (including property rights) and contributes to efficiently functioning markets. 6 From the point of view of poverty transitions, how power is exercised at the community level is considered critical. The extent of transparency, participation, accountability and local organizing in short, empowerment has relevance not only for competitive and inclusive politicking and sound public sector management, but also for realizing economic and social mobility. The relevant power consists of at least three elements: power to influence political decision making, power to take economic actions (power over resources), and self-power or confidence. The power to influence political 6 See, for example, Keefer and Khemani

11 decision making can be manifested by voting in elections, participating in public meetings and discussion fora, direct actions to contact politicians and people s assessments of whether politicians are responsive and accountable to them. The power to take economic actions refers to the ability to buy and sell land and other assets, to set up a business, and to participate freely in economic markets without facing discrimination or other barriers to access. Finally, self-power or confidence is brought about by a cultural and educational environment that is inclusive and does not hold back people on the basis of gender, social identity, ethnicity, or other factors. The hypothesis is that people who are empowered at each of these three levels will find it easier to take the necessary economic decisions and actions to escape poverty. Capturing empirically these different elements of power is not an easy task and the study will rely on a combination of information from the household survey and the qualitative data collection at the community level to construct different measures of empowerment for the purpose of testing the hypotheses. In addition to focusing on the relation between empowerment and poverty reduction, the study will also pay attention to the time sequencing of shifts in power in its qualitative exploration, i.e. do social power relations or changing power relations through collective action open up new opportunities? Differences in the composition and status of social groups will also be an important dimension of this analysis. Finally, freedom and aspirations constitute individual attributes acquired in environments of national and local democracy and empowerment. While these attributes may have more to do with perceptions of reality than with objectively measurable dimensions, they reflect basic motivations and aspirations that contribute to attempts to 11

12 escape poverty. We have no preconceived notions about which aspects of freedom will turn out to be the most important in this regard, and the study will thus collect information on people's own definitions of freedom, and the multiple aspects of freedom, such as freedom of speech, religion, to form associations, etc. However, given the study's focus on poverty transitions, we will limit the inquiry on aspirations to people's economic aspirations. In summary, the basic political hypothesis that will be tested by the study is that households have a higher probability to escape poverty on a permanent basis if they (a) live in communities with democratic and accountable local governments, (b) are empowered and able to take a number of economic and political actions that increase their control over their lives, and (c) enjoy a wide range of freedoms and have positive economic aspirations towards the future. Sampling Country selection will be based on four characteristics deemed to be particularly important as determinants of the extent of poverty transitions and of the way in which these transitions occur. They are level of income, rate of growth over the last decade, extent of political rights and civil liberties as a proxy for democracy, and extent of openness to global markets. Countries will be selected to cover all the regions. At the country level, the study will be based on panel data where possible, and at a minimum this will include 4-to-5 7 of the 20 countries participating in the study. The challenge of the study is to have a measure of a household s poverty status for at least two points in time, separated by a reasonable interval. Unfortunately, such data may be 7 This is a conservative estimate. We expect to be able to tap into panel data in several countries. 12

13 readily available only for a handful of countries for which panel data studies have been conducted. While every attempt will be made to utilize panel data, in countries where this is not possible the study will make use of an existing or ongoing household survey that includes expenditure data. These studies will include 30 randomly chosen communities in high and low growth areas of the countries. Within communities, the principle of randomization will also be maintained in selecting households. Given that a ten-year recall of consumption expenditures is not possible, the study will gather information using four alternative poverty measures: expenditure data, household assets, food security, and self- assessments of current and previous status through a Ladder of Life ranking. First, if the new survey can be linked to a prior survey that did collect data on consumption expenditures, the data from the earlier survey will be used to compute the initial poverty status. Second, information on assets will be collected, for instance, on ownership of major consumer durables, the number of rooms in the house, and so forth. From this data, an index of asset ownership will be constructed to provide a basis for determining the magnitude of change across two points in time. The third approach will be to focus on a simple measure of poverty, the access to food, as measured by the number of meals per day a household is able to afford. In particular, households will be asked how likely they are to be forced to skip meals (because they cannot afford them) today, and also in the initial period. A household that used to be forced to miss meals, but no longer has to do so will be considered to have 13

14 undergone a poverty transition. The question will be phrased so as to account for seasonality, and the respondent will be asked to focus on a particular time in the year such as the months before a harvest, or the period following a festival. A fourth approach will use a Ladder of Economic Mobility. Each household will be asked to rate its status on a 10 point scale in the current and initial periods. It will also be asked to define a rung on the ladder corresponding to the poverty line. A household which reports that it was below this rung in the initial period and above this rung now will be considered to have emerged from poverty. The reliability of this method of assessing poverty status will be evaluated by checking the consistency of information obtained from the ladder exercise with that obtained from the consumption expenditure survey. It will be possible to perform this check for the current poverty status for all the countries, and for initial poverty status only when a prior survey is available. Data Collection Instruments Research teams will begin preparations for data collection with desk reviews of current major analytical works on poverty in their countries, including important panel or cross sectional studies, poverty maps and other social science research. These will be used to inform the sampling frame, data collection, and analysis of findings. Other preparatory work will involve interviews with national policymakers to elicit their views on key macro events and policies affecting poverty over the last ten years, and their perceptions of the principal factors contributing to escaping poverty in the country. 14

15 Primary data collection tools include a questionnaire and five qualitative instruments. 8 The field instruments are being revised following pilot tests in the fall of 2003 in Ethiopia, India, Peru, The Philippines and Romania, and a workshop in December 2003 to discuss the pilot methods and findings. They instruments include: A community profile to collect basic demographic, economic, social and political data on the communities under study. A questionnaire with modules on Economic Characteristics of Households (Assets, Sources of Cash Income, Household Expenditures, Credit and Debt, Access to Employment and Markets); Education and Health; Social Capital (Groups and Networks; Trust and Solidarity; Collective Action and Cooperation; Social Cohesion); Power, Governance and Access to Information; Freedom, Crime and Violence; Perceptions of Well-being and Aspirations. Semi-structured household interviews to gather life stories, including views and experiences related to key events and processes affecting that household s transition over the previous ten years and aspirations for the future. Two focus group exercises to explore i) community level factors affecting wellbeing and mobility; and ii) people s understandings of freedom, power, and democracy and how these concepts relate to mobility; and their perceptions of the functioning of local level democracy, elected officials and government functionaries, community leaders, local services and the media-- and how the functioning of these institutions affects mobility. Both exercises will also examine aspirations. Key informant interviews and mini case studies with policymakers and community leaders to collect information for key event timelines and to explore factors that have had widespread impacts on mobility. The fieldwork will engage a variety of informants. The questionnaire and semistructured interviews will be conducted with all households in the study. Focus groups will be held with men, women and youth in all of the study communities. Key informant interviews will be held with leading poverty specialists and policymakers as mentioned previously, and in the study communities, with elected officials, public functionaries, and community leaders. 8 Draft and pilot versions of the instruments are available upon request. These will be finalized shortly for the full study. 15

16 Participatory research approaches have demonstrated that local people have indepth knowledge and strong analytic skills to contribute to poverty research. To maximize such input, research teams will ensure that local people: are informed of the study purpose and tools; can explore and analyze a range of topics of importance to them; and can provide feedback on and validate findings. Such efforts to meaningfully engage local people in the research process are designed not only to improve the quality of information collected, but more importantly, to empower poor or disadvantaged groups to use the study to support their own planning and action. Funds are being sought to support local follow-up activities. Data Analysis Quantitative Analysis. The first goal is to produce a rich descriptive picture of the evolution of poverty status. Examples of questions that will be explored include: Is there an association between emergence from poverty and a household s social capital? Are more educated households more likely to emerge from poverty? Are changes in the level of crime or insecurity associated with changes in poverty status? Is there a correlation between a household s perception of freedom and changes in its economic status? This will be the first study to report such findings for a substantial number of countries. A more ambitious, and considerably more difficult, task will be to try to establish causality: what factors contribute to emergence from poverty? For example, the data collection will enable comparisons among the poverty transitions of households with different initial levels of social capital. Relationships among explanatory variables measured at the community level will also be explored, such as the importance to household outcomes of growth or quality of governance at the local level. 16

17 The use of pocket PCs which enable enumerators to enter data directly during interviews is being explored. This technology has been used successfully in a number of household surveys, and is particularly indicated for short and medium-length questionnaires with a relatively simple structure. This makes it possible not only to bypass completely the data entry phase, but also to introduce enhanced quality control in the field by virtue of the fact that range and consistency checks are done instantaneously. After data collection is completed, the data are downloaded directly into a personal computer that has been preprogrammed to produce basic descriptive statistics for all the questions as well as a set of simple cross-tabulations. These tabulations will provide the researchers with a very rapid insight into some of the major findings at the household level. Qualitative Analysis. The most often cited benefits of qualitative poverty research lie in its ability to uncover a more multidimensional understanding of poverty and wellbeing, to produce more accurate and complete information, and to enable discovery and exploration of the unknown. More generally, qualitative research is valued for the local groundedness and richness and holism of results. When linked to quantitative studies, qualitative work often performs key roles of signaling conceptual areas and hypotheses for testing, and helping to explain or interpret quantitative findings. It is also helpful in understanding people s own definitions and cognitive maps of concepts. Triangulation will be the cornerstone of data analysis, with cross referencing of the data carried out from the qualitative instruments across the households and sites visited, and with secondary data. Once field work is complete, documentation of primary data will be completed, and systematic content analysis will be conducted on the 17

18 documents with the aid of QSR NUDIST to support data coding and analysis. The qualitative work will also entail some numerical data collection. The instruments involve rankings, scorings, and ratings to facilitate shared analysis during focus groups. Integration of Quantitative and Qualitative Analyses. A key task for this project is to integrate the data from the household surveys with the information obtained from the focus groups and the open-ended individual interviews. The information from focus groups and in-depth household interviews on mobility and the key study concepts will be valuable in their own rights. In addition these findings will enrich our understanding of the quantitative data. For instance, the nature of social capital, and the reasons why it may be high or low in a community can only be understood from the focus group discussions. Or, if the econometric results show that membership of an indigenous group adversely affects the probability of exiting poverty, the in-depth interviews may clarify exactly what kinds of barriers indigenous groups face. The qualitative work can also help in the understanding of outliers. Also important, a number of community level variables that are used in the regression models/tables will have to be provided by the focus groups. The final stage of the analysis will be to develop a synthetic cross-country view of the results. In line with the factors that contributed to the selection of countries, the focus will be on economic performance (level of income, growth rate), democracy and governance, and degree of exposure to the global market. Although it will not be possible to prove formally the role of these factors in determining poverty transitions, the view across countries will aim to detect common patterns of poverty transitions and their determinants in countries that have similar economic, social, and political conditions. 18

19 The global synthesis will also enable some reflection on patterns of findings across countries associated with low and high growth at the local level and with other community-level characteristics. Work Program The research design phase of the study will be completed in January 2004 following five international methodology workshops and five country pilots of the methodology. Research partnerships will then be mobilized in 20 countries to reflect the methodology criteria discussed above. To gauge interest, contributions by the research institutes or local Bank offices undertaking the studies will be required. The country level data analysis is expected to be completed by December 2004, after which the national reports will be shared at country-level workshops and at an international workshop. The global synthesis phase will then be launched, which will require one year and will conclude with a multidisciplinary international workshop to share findings. Outreach Country-level dissemination and follow-up activities will be formulated in close coordination with national Learning Groups to foster capacity building and ensure strong links with policymaking processes, including PRSP activities where relevant. In some contexts, such as Uganda and Vietnam, participatory research has become institutionalized and well integrated into national policy processes. To support the flowering of such processes elsewhere, Moving Out of Poverty Learning Groups will be established in each of the study countries. The Groups will include a small number of national and local government leaders, the study team leaders from the local research institutes, selected policy-oriented civil society organizations, private sector 19

20 representatives, and Bank staff and other donors. The Groups members will provide expert input into study preparations (e.g., adapting instruments to local contexts) to ensure strong policy relevance. They will also be invited to participate in training of field teams, fieldwork, and local feedback activities. The members will especially be expected to review early study findings, and once the report is complete, to champion dissemination activities and policy follow-up in their countries. All country reports will be published in 2005 and the global reports in Country level workshops and media activities will also be conducted. It is hoped that these partnerships might lead to improved capacities for participatory work and use of poverty diagnostics for policymaking, as well as action at the local level where the research takes place. Should funding become available to support local follow-up action in participating communities, it is expected that the Learning Groups could help to program and oversee the funds. Proposed Three-Year Budget Activities 1. Finalization of Data Collection Instruments 2. Identification and Contracting of Country Research Institutes 3. Training, Technical Support and Supervision of Research Teams in Qualitative and Quantitative Tools 4. Software Development 5. Country Fieldwork in 20 countries 6. Technical Support for Country Data Analysis and Report Preparation 7. Publication and Dissemination of Country Studies 8. Analysis and Publication of Global Findings 20

21 Budget Calendar Years Amount (US$) 20 Country Studies 2,000,000 Study Coordination 300,000 Administration and Research Analysis 200,000 Specialist Consultant Support for Training, Technical Support and Supervision 100,000 International Travel (12 trips) 100,000 Software Development 100,000 Global Synthesis 200,000 TOTAL $3,000,000 21

22 Annex Literature Review, Summary Tables and Bibliography Literature Review What We Know. We know that sustained and high growth over a long period of time is good for the poor. We know that under different initial conditions, particularly in inequality, growth has a different poverty reduction impact. We know that there are great variations across countries, and that within countries there are lagging regions. A review of the economic literature based on household surveys, cross-country regressions, and panel studies confirms that the cluster of factors that are most often associated with growth that benefits the poor are: household endowments; geography; economy-wide factors including macroeconomic volatility and trade policies; and initial levels of inequality in assets, land, and education and literacy. Recent studies using panel data and poverty maps over time contribute further to our understanding of sources of growth and the role of inequality in initial conditions. What We Don't Know. We don't know what policy sequencing and design works best in developing countries in different historical, political, social and economic contexts to achieve growth that benefits the poor. While there is agreement on broad actions needed to spur growth, this is of limited use as these broad principles have not brought about growth in many developing countries. There is agreement that we don't understand enough of the political economy of change and that we don't know the mechanisms needed at the local level to support increasing economic choice and freedom. We also don't know how and why some people move out of poverty and stay out of poverty while 22

23 others in the same area fall into poverty. We don't know if there are significant gender differences in strategies out of poverty. This study seeks to include factors that have not been a systematic part of the analysis of either poverty reduction or growth. The additional factors are: economic aspirations, social exclusion; social networks and norms, social capital; quality of both national and local democracy and governance; and perceptions of freedom and power. Mobility Panel Studies Panel data sets have been used to study pro-poor growth, poverty dynamics, chronic and transitory poverty and vulnerability. While most studies use monetary measures of well-being such as income or expenditure, selected panel data investigate subjective well-being (for example Russian and Peruvian surveys). Panel studies also vary in the frequency and duration of data collection (see Tables 1 and 2 of Annex I); this study will be particularly interested in more recent panels with a minimum of five-year intervals. Existing literature on panel surveys in developing countries illustrates that a large percentage of poor households move in and out of poverty while only a minority of them remain persistently poor (Baulch and Hoddinott 2000). Studies by Jalan and Ravallion (2001) and Baulch and McCulloch (2001) highlight the importance of this distinction between chronically and transient poor in designing effective poverty reduction policies. Baulch and Hoddinott (2001) argue that transitory component of poverty reflects inability to cope with negative shocks and the related inability to smooth consumption among 23

24 these households. 9 The smaller chronic component of poverty is associated with low endowments of assets and inability to translate these into more adequate income. McKay and Lawson (2003) point out that while panel data may illustrate that the chronically poor group have distinctive characteristics (lack of assets or high dependency rates), there remains a need to develop non-monetary dimensions of chronic poverty to unravel the underlying characteristics in more detail. Mobility Other Quantitative and Qualitative Studies Panel data are costly, time-consuming, and can be weakened by significant attrition rates in the sample population. Some quantitative and qualitative studies instead incorporate retrospective and often more open-ended methods to explore poverty dynamics. Tables 5 and 6 of Annex I provide findings from these approaches. The value of these studies is the incorporation of a broader set of social, cultural, economic and organizational factors to expand our understanding of the mechanisms behind household transitions in and out of poverty and enduring poverty. Scott (2000) applied anthropological life history interview techniques to study the declining fortunes of households with small farms over an eighteen-year period in Chile. He finds that roughly half of the households fell into poverty due to transitory factors such as livestock lost to drought and diseases; the remaining were affected by more permanent processes of impoverishment, such as unprofitable investment projects, high dependency ratio and political discrimination in the labor market. 9 A study by Dercon and Krishnan (2001) explores seasonal factors affecting transient poverty, and finds large variability in poverty as households suffer from a variety of seasonal shocks related to weather and more idiosyncratic factors associated with crops, livestock and illness. 24

25 In another field study, Krishna (2002) argues that longitudinal studies pay little attention to what households are doing themselves to deal with poverty in their midst. In his fieldwork in 35 villages in the state of Rajasthan, India, instead of using the classic questions to measure expenditure, he asked the villagers to determine collectively what constitutes a state of poverty. More than 85 percent of all cases of decline into poverty included a combination of health problems, expenses on health care, debt from large and unexpected expenditures and social expenditures. By contrast, diversification of income sources played a major role in escaping poverty. A recent 30 year follow up study of favelas in Rio de Janerio, Brazil is examining relationships between life history patterns and macro political and economic changes (including local, national and global and their effects), and testing the mediating effects of civil society and social networks (Perlman 2003). Study findings highlight issues of fear, insecurity, exclusion and the unfulfilled promise of democracy as important factors in preventing economic advancement. The mixed methods research is tracing life histories across 4 generations to explore intra- and inter-generational poverty dynamics. Social Groups and Social Identity The importance of social groups and social identity to development outcomes is well established. With some notable exceptions, women in developing countries face diverse disadvantages compared to men: they are less educated, own fewer assets, have lower earnings, are more vulnerable to shocks, and frequently exercise less voice at home and in public life. In addition, caste, race, ethnicity, and religion can be sources of enduring social barriers. Lower educational attainments and income, for instance, 25

26 frequently coincide with being indigenous in Latin America (Patrinos 2000). In Voices of the Poor, in fact, people identified quite diverse triggers for ongoing and sometimes severely discriminating and degrading treatment, including having low status work, residing in a bad neighborhood, having a poor appearance or darker skin, having a disability or HIV-AIDS, or even the mere fact of being poor. Frequently, these disadvantages overlap. In explaining the persistence of inequalities at the micro level, Heller and Mahoney (2003, 22-26) stress the influences of social identity and differences in status and power among social groups which shape their relations. These differences, which operate independently of the actions of any given individual, help more powerful actors to sustain their privileges and distinctions, and to effectively deploy mechanisms of exploitation and opportunity hoarding. Similarly, Tilly argues that differences among social groups deserve special attention because they provide clearer evidence for the operation of durable inequality, because their boundaries do crucial organization work, and because categorical differences actually account for much of what ordinary observers take to be results of variation in individual talent or effort. (1999, 6) In addition to factors related to social status and position, the possibilities for collective action powerfully affect development prospects. Since Making Democracy Work 10 there has been an explosion of empirical work exploring the links between social capital, and incomes, growth, trade, governance and political performance. Social capital can broadly be referred to as the norms and networks that enable collective action. As with economic factors, the effects of social factors on households' ability to escape poverty operate at the macro level and the meso/micro level. At the macro level, social 10 Putnam,

27 capital includes nationwide institutions such as the political regime, the rule of law, the court system, national solidarity norms as reflected in social security systems, etc. Two components of social capital social cohesion and trust have been found to be critical qualities for reducing corruption, reinforcing government performance, and providing a conducive environment for reform. Knack and Keefer (1997) have shown a significant relationship between measures of trust and civic cooperation and economic growth. Rodrik (1999) has shown that countries with the steepest falls in growth after exogenous shocks after 1975 were those that were socially divided in terms of income inequality, ethnic and linguistic fragmentation or conflict, low levels of social trust; these countries also had weak institutions for managing conflict. 11 La Porta and others (1997) have found positive relationships between trust and judicial efficiency, anti-corruption, bureaucratic quality, tax compliance, and civic participation. In addition, Ritzen, Easterly, and Woolcock (2000), using a wide variety of data and measures of trust, income inequality, ethnic heterogeneity, voice and accountability, and others, find that social cohesion and inclusiveness are critical for generating the trust needed to implement reforms. 12 The contribution of social capital to improved development outcomes is perhaps most evident at the individual and community level. In a household survey in rural Tanzania, Narayan and Pritchett (1997) show that higher village-level social capital as measured by membership in groups with particular characteristics generated higher household incomes. In such disparate places as Bolivia, Burkina Faso, and Indonesia, household surveys measuring social capital have found positive associations with 11 Cited in Narayan Cited in Narayan

28 household consumption, asset accumulation, and access to credit. 13 Social capital also has important positive impacts on business performance. Studies of agricultural producers and traders in Madagascar show that social relationships are more important than input prices. These ties are used to lower transactions costs of exchange; and ties to creditors are critical during times of financial hardship. 14 Evidence from community-driven development projects and other research have shown that social capital can have an important influence on the delivery, performance, maintenance, and sustainability of public services. King and Ozler (2001), for example, find that that primary school teachers in Nicaragua are better able to increase test scores when PTAs are given more decision making power. In the Philippines, Jimenez and Paqueo (1996) find that schools that rely more on community contributions use their resources more efficiently. 15 Similarly, analyses of water projects in Indonesia, India, and Sri Lanka find that greater community participation is associated with better water supply. Heterogeneity in project effectiveness is largely explained by the ability of a community to engage in collective action, and high levels of social capital improve participation in design and monitoring. 16 In another study, econometric analysis of 121 rural water supply projects found organizational capacity at the local level to be critical in sustainable functioning of water systems. 17 Finally, a study of water supply projects in Cote d Ivoire found that when responsibility was shifted from central governments to water user groups, breakdown rates were reduced from 50 to 11 percent, and costs fell by 13 Grootaert 1999; Grootaert and Narayan 2000; Grootaert, Oh, and Swamy Cited in World Bank Fafchamps and Minten Cited in World Bank This section draws heavily from Mansuri and Rao Isham and Kahkonen 1999a and 1999b. Cited in Mansuri and Rao Narayan

29 nearly 70 percent. Importantly, these results were sustained only where well-functioning community organizations existed. 18 But are strong social ties always pro-poor? And which networks most help poor people escape poverty? On both of these questions, the social capital literature is making important strides, and the global reach of this study can continue to shed light on the forms of social capital which may be most critical for movement out of poverty. Social capital which hinders development include communities or networks which are isolated, parochial, or working at cross-purposes to society s collective interests (e.g. gangs, drug cartels) (Woolcock and Narayan 2000) And associations which help poor people survive daily life which are typically very local and consisting of people of similar socio-economic status are often not as valuable for escaping poverty as the ties which cross social divides. These latter networks, which can extend beyond community borders and encompass groups of varied social and economic status, are found in diverse contexts to be invaluable springboards for economic advance and fostering social cohesion. However, poor people often find themselves excluded from many such valuable associations. For these reasons, both the questionnaire and the qualitative work feature detailed questions about the composition, reach and impacts of local networks. The questionnaire and qualitative instrument also gather data on levels of crime and violence and how they may affect social cohesion in communities. Where there are strong social bonds and high levels of trust, neighbors can help to police their communities and reduce levels of crime. When these ties are weak or break down and violence and crime are pervasive, society s health and economy suffer and fear and distrust often undermine the associational life of communities (Moser and Holland 1997). 18 Hino Cited in Narayan