Analyzing Urban Poverty: A Summary of Methods and Approaches

In recent years an extensive body of literature has emerged on the definition, measurement, and analysis of poverty. Much of this literature focuses on analyzing poverty at the national level, or spatial disaggregation by general categories of urban or rural areas, with adjustments made for regional price differentials. Yet for an individual city attempting to tackle the problems of urban poverty, this level of aggregation is not sufficient for answering specific questions such as where the poor are located in the city, whether there are differences between poor areas, if access to services varies by subgroup, whether specific programs are reaching the poorest, and how to design effective poverty reduction programs and policies. Answering these questions is critical, particularly for large, sprawling cities with highly diverse populations and growing problems of urban poverty. Understanding urban poverty presents a set of issues distinct from general poverty analysis and thus may require additional tools and techniques. Baker and Schuler summarize the main issues in conducting urban poverty analysis, with a focus on presenting a sample of case studies from urban areas that were implemented by a number of different agencies using a range of analytical approaches for studying urban poverty. Specific conclusions regarding design and analysis, data, timing, cost, and implementation issues are discussed. This paper - a product of the Urban Unit, Transport and Urban Development Department - is part of a larger effort in the department to promote strategies for reducing urban poverty.


Introduction
In recent years an extensive body of literature has emerged on the definition, measurement and analysis of poverty. 1 Much of this literature focuses on analyzing poverty at the national level, or spatial disaggregation by general categories of urban or rural areas with adjustments made for regional price differentials. Yet for an individual city attempting to tackle the problems of urban poverty, this level of aggregation is not sufficient for answering specific questions such as where the poor are located in the city, whether there are differences between poor areas, if access to services varies by subgroup, whether specific programs are reaching the poorest, and how to design effective poverty reduction programs and policies. Answering these questions is critical, particularly for large, sprawling cities with highly diverse populations and growing problems of urban poverty. 2 Understanding urban poverty presents a set of issues distinct from general poverty analysis and thus may require additional tools and techniques. This paper attempts to summarize the main issues in conducting urban poverty analysis, with a focus on presenting a sample of case studies from urban areas that were implemented by a number of different agencies, using a range of analytical approaches for studying urban poverty. A complementary document, The City Poverty Assessment: A Primer (Hentschel and Seshagir, 2000) provides a more in-depth discussion of specific types of analysis. Section I discusses the approaches to urban poverty analysis, Section II presents a summary of the case studies, and Section III concludes.

Designing Urban Poverty Studies
For the purposes of this paper, urban poverty analysis (UPA) is defined as the process of gathering, analyzing, and presenting information on the extent, location, and conditions of poverty in a given city. This can then be used to generate a city poverty profile that policy makers, community members and academics can use in answering questions about urban poverty and to identify appropriate responses. It also provides a baseline from which changes in policies, investments and activities may be measured. As the conditions and quality of poverty are not static, a good city poverty profile should serve as a dynamic starting point for better understanding poverty problems and appropriate responses.
While there is no single approach to conducting urban poverty assessments, there are some common good practices that may facilitate the process of thinking through the design of a city poverty profile, which are discussed below.

a. Urban Context
What issues are specific to urban poverty?
While the dimensions of poverty are many, there is a subset of characteristics that are more pronounced for the poor in urban areas and may require specific analysis: 3 commoditization (reliance on the cash economy); overcrowded living conditions (slums); environmental hazard (stemming from density and hazardous location of settlements, and exposure to multiple pollutants); social fragmentation (lack of community and inter-household mechanisms for social security, relative to those in rural areas); crime and violence; traffic accidents; and natural disasters.

b. Definitions and Identification
Who are the poor? What poverty indicators will be used in the city poverty profile?
The application of quantitative poverty measures to a city should be viewed as the start of a city profile; distinguishing the poor from the non-poor (and the relative sizes of these groups), identifying their characteristics, facilitating the process of identifying the location of the poor, and highlighting groups that may be particularly vulnerable to poverty (i.e., female-headed households, migrants, youth, etc.). This basic data can be used to inform the selection of more indepth qualitative measures.

Box 1: Approaches to Measuring Urban Poverty
Poverty is multidimensional, thus measuring it presents a number of challenges. Beyond low income, there is low human, social and financial capital. The most common approach to measuring poverty is quantitative, money-metric measures which use income or consumption to assess whether a household can afford to purchase a basic basket of goods at a given point in time. The basket ideally reflects local tastes, and adjusts for spatial price differentials across regions and urban or rural areas in a given country. Money-metric methods are widely used because they are objective, can be used as the basis for a range of socio-economic variables, and it is possible to adjust for differences between households, and intrahousehold inequalities.
• Participatory methods: This typically relies on qualitative approaches to capture aspects of urban poverty that may not be identified through pre-coded surveys. Through tools such as focus group discussions, case studies, and individual open-ended interviews, it is possible to determine the perceptions of poverty, identify priority needs and concerns, and gain insight on the effectiveness of programs and policies from the perspective of the beneficiaries.

c. Location
Where do the poor live?
A city poverty profile is by definition a spatial tool, defining the extent and nature of poverty within a given area. Providing an accurate mapping of where the poor live and are concentrated within cities is an important input required for targeting appropriate responses. Spatial information on land use patterns, slum location, and physical location of facilities through Geographic Information Systems (GIS) are also critical inputs. In many cases, especially when the poor live in informal settlements, they may not be included in existing administrative data.

d. Access and Accessibility
How does the city (policies, environment, and infrastructure) mitigate or contribute to poverty?
In providing its citizens with infrastructure (roads, transportation, water, electricity, etc.), education, health care, legal, political, and financial institutions, and employment opportunities, the city provides the tools that individuals use to generate their livelihoods and well-being. Illustrating the availability and accessibility of these municipal services to the poor, should be a key component of a city poverty profile. This can also identify areas that are particularly vulnerable to social exclusion due to the lack of services.

e. Characteristics, Opportunity, and Constraints
What is the nature of poverty?
While income-based poverty measures provide a fair sense of which part of the population may have unmet needs and where they are located, these measures fail to capture the dynamic aspects of poverty, in terms of the cause and extent of deprivation, risk factors, and the coping strategies employed. This can include analyzing vulnerability, urban-rural linkages, and perceptions. Qualitative methods are often used for this kind of analysis.

Box 2: A standard poverty profile for cities
Any profile of urban poverty would be based on a set of core information as described below.
Household surveys/census: In general this would include the following information disaggregated by income group (e.g. quintile): Location (within the city) Household size, structure Demographics Education levels Household expenditure patterns Employment (status, occupation, hours worked) Housing characteristics (tenure status, physical condition) Access/quality/affordability to: infrastructure (water, sewage, energy); health care; education; social services.
Administrative data: This would include data collected by various public agencies, ideally disaggregated by geographic areas within a city.
Municipal spending by sector, location Infrastructure (roads, public standpipes, schools, hospitals) Health and nutritional status Education outcomes Crime and violence statistics

Using Urban Poverty Analysis
To justify the expense (in time and resources) of conducting city level poverty research it is important to have a clear link between the information to be collected and the application of that information in city policies, programs, etc. A common pitfall in conducting research is a lack of clarity in objectives, resulting in the collection of unclear and sometimes useless information. 7

a. Strategy Development
In the process of balancing the dual objectives of providing quality of life and creating competitive cities, many local authorities are engaging in urban planning and city-wide strategy development activities. Urban poverty analysis can provide valuable inputs into the process of city strategy development.
City Development Strategies (CDS), as supported by Cities Alliance 8 , back a 'comprehensive analysis of key urban issues in a particular city, and help define strategic options and implementation alternatives.' 9 Urban poverty analysis can facilitate the identification of 'key urban issues' through quantitative measures of urban poverty, and qualitative measures of community priorities. Local Economic Development (LED) 10 is a more economically driven city strategy tool that 'offers local government, the private sector, the not-for-profit sectors and the local community the opportunity to work together to improve the local economy. It focuses on enhancing competitiveness, and thus increasing sustainable growth; and also on ensuring that the growth is inclusive.' Urban poverty analysis can also be used to highlight opportunities and constraints to economic opportunities (e.g. quantitative focus on employment structure, spatial location of workers, and qualitative analysis of household responses to poverty, labor-industry relations).

b. Service Delivery and City Budgeting
Service delivery is a core function and often a central challenge for local authorities. The gap between the expectations of constituents and the capacity of municipalities to deliver is defined by a range of financial, political and management issues. Increasingly, local authorities are required to explore different approaches to service delivery, particularly in meeting the needs of the poor 11 . New approaches to service delivery also require new types of information. Urban poverty analysis can be an effective tool in gathering and analyzing information for the purposes of designing and implementing mechanisms for service delivery.
Slum Upgrading 12 Infrastructure Tenure, Land administration City budgeting

c. Policy and Project Monitoring
Without community buy-in and accountable implementation mechanisms, even the most enlightened of urban projects and service delivery mechanisms will fail. Monitoring the impact or effectiveness of a project over time can be a critical mechanism for both involving communities and learning from mistakes (and avoiding them in future projects). Integrating urban poverty analysis into World Bank urban projects may be useful both in designing projects (utilizing data findings) and in monitoring their impact. Often projects are designed to target specific groups, but without baseline data it is difficult to measure the impact of targeting measures, and to learn from successes or failures.
Urban poverty analysis that uses available census and administrative data and specially designed surveys can provide an overview of the change in people's quality of life over time in areas such as income, assets, health outcomes, and education outcomes. This general picture can also be tailored to highlight the impacts of specific investments (slum upgrading, infrastructure investments etc.), to determine their quality, cost-effectiveness and possible replicability; and to provide a de facto baseline for any future actions and investments. In this way, urban poverty analysis is a valuable tool for urban policy development, providing both an input (specific local information) as well as a project output (a starting point for a project baseline). 11 See World Development Report 2004, Making Services Work for the Poor. World Bank. http://econ.worldbank.org/wdr/wdr2004/text-30023/ 12 This may include improving and/or installing basic infrastructure like water, sanitation, waste collection, access roads and footpaths, storm drainage, lighting, public telephones, etc. Upgrading also deals with regularizing security of land tenure and housing improvements, as well as improving access to social support programs (e.g. health, education) and municipal services (e.g. water, sanitation, waste collection, storm drainage, street lighting, paved footpaths, roads for emergency access). http://www.worldbank.org/urban/upgrading/index.html

Box 3: Poverty Mapping in Urban Areas: The Case of Brazil
In response to a growing demand for disaggregated data, Brazil has in recent years been compiling and updating an Atlas of Human Development. The initiative, being supported by UNDP, is being implemented by two think tanks that have developed a set of human development indicators using census data for the past four decades. The indicators were based on 38 variables in five sectors; education (literacy and enrollment rates), health (life expectancy), housing (water supply and sewage facilities), income (inequality and population under the poverty line), and infant mortality.
The most recent version of the Atlas has attempted to cover 3 large cities (Belo Horizonte, Rio de Janeiro and Recife) to look at poverty within cities. The data has been made widely available allowing researchers to further analyze information. It has been used for targeting social programs. It is also being used by members of the private sector, such as Telemar, a telecommunication provider which as used the Rio de Janeiro Neighborhood Atlas to market differentiated telephone services to poor areas within the city.
Source: Experience with the Development and Use of Poverty Maps, Case Study Note for Brazil, in Henninger, N. and Snel, M.

Data for Urban Poverty Analysis
The shape of any urban poverty analysis will be largely determined by the questions being asked by those designing the profile, the extent and quality of available data, and the amount of time and money that can be committed to the development of a profile. Within these parameters, there is a range of tools that can be employed to gather data for a city poverty profile.

a. Population census
The population census contains basic information about all citizens in a country. While usually only conducted once a decade, the census may provide valuable information about housing and basic service access, education levels, and employment. Since the census is by definition a counting of everyone, the data can be disaggregated to the city, municipal, and neighborhood level. While the data do not include indicators of income or consumption, proxies for welfare can be developed using access to basic services (e.g., UBN). In addition, a relatively new approach has been developed that combines census and household survey data to generate poverty and inequality measures at disaggregated levels. This approach is very powerful for urban poverty analysis. 13

b. Administrative data
Many local authorities collect significant amounts of information about their cities, but often that information is gathered and stored in ways that make them difficult to use or access. In developing a city poverty profile, existing administrative data can prove to be a rich resource of information. Such data may include: Information on location of facilities such as schools, hospitals, public standpipes, etc. Costs and expenditures by sector and function, Tax revenue by source, and Staffing statistics, etc.

c. Household surveys
Household surveys offer a more in-depth understanding of living conditions by expanding the amount of information gathered from households, but carefully limiting the number of households surveyed to create a representative sample of the population. Household surveys are used both nationally and at the city level.
National household surveys that may provide insight into urban poverty include: 1. Employment Surveys: information in employment and unemployment patterns and fluctuations. These also include information on household income, housing features and household demographics. 2. Demographic and Health Surveys 14 : are designed specifically to explore the incidence of disease and use of health facilities. 3. Living Standard and Measurement Surveys 15 : collect household data that can be used to assess household welfare, to understand household behavior, and to evaluate the effect of various government policies on the living conditions of the population. LSMS surveys collect data on many dimensions of household well-being, including consumption, income, savings, employment, health, education, fertility, nutrition, housing and migration. 16 At the city-level the following types of household surveys may be developed to explore particular issues.
Service Satisfaction and Needs Surveys 17 : ask city residents in-depth questions about the quality of services they receive and their needs. These will likely be tailored to address specific services, and may be used to collect data over time. Multi-topic City Surveys 18 : are tailor made to city needs and may collect both qualitative and quantitative data. Quantitative data primarily includes household welfare (i.e. income, assets), service access, health service utilization, public transport use etc. Qualitative data may highlight service satisfaction and perception surveys. Developing a multi-topic city survey is likely the most comprehensive way to analyze the dynamics of urban poverty in a given city.

d. Participatory assessments
Participatory assessments go beyond household surveys to gather more qualitative data on individual and community perceptions. There are a range of tools 19 19 Many of the tools of participatory assessments fall under the rubric of 'participatory rapid appraisal' or PRA. While many of these techniques originated within the study of rural poverty, they have been increasingly applied to the urban context. See Annex 1: Case Summaries 4, 7, & 8.
conducting participatory assessments, including focus groups, community meetings, community mapping, transect walks, and in-depth interviews. Participatory assessments may be used to facilitate the selection of indicators for household surveys and/or they may be used to explore particular issues that emerge through surveys in greater depth 20 .

e. Geographic Information Systems (GIS)
GIS data use geographic location as a reference for each database record and can be very powerful within cities for identifying spatial growth patterns, slum locations, access to public infrastructure, and land use patterns. This information can be combined with census and other data to determine the spatial dimensions of poverty and access within a city.

Urban Poverty Analysis in Practice
Short summaries of a number of urban poverty studies have been included in Annex I (see Box 4), with an emphasis on the specifics of methodology and data rather than specific outcomes and analysis. The studies have been selected because they represent a number of approaches to urban poverty analysis and offer a range of geographic diversity. There are no clear 'best practices' regarding urban poverty analysis as a city poverty profile is most fairly judged in relation to its study objectives.
From the case studies, a number of examples have been highlighted for their strengths in particular elements of urban poverty analysis. Below is a listing of several key themes and the studies that cover them well. This is intended to serve as a summary list and does not reflect the depth and subtleties of the studies. The number in brackets refers to the case summary that can be found in Annex 1.

Developing a profile of City Poverty
In several of the studies, the objective has been to develop an overview of the conditions, extent, implications, and activities relating to poverty within a particular city. These may be carried out to complement national poverty assessments or to provide input into city development strategies or specific urban projects/ investments.

Geographic and other Disaggregations of Poverty
Although urban poverty analysis is implicitly spatial, it cannot be assumed that all studies focus on the geographic disaggregation of poverty. Within densely populated cities it becomes increasingly challenging to target services and attention at the poor who often live in close proximity to their better-off neighbors. Urban poverty analysis may also disaggregate data to examine households, social risk, violence and other characteristics of poverty.

Understanding the dimensions and characteristics of living in poverty
Traditionally the bulk of poverty analysis has been on rural areas, and so there is still much to be learned about the characteristics and strategies in response to urban poverty. An understanding of the specific characteristics and dimensions of urban poverty may lead to more equitable and cost-effective policies.

Linking poverty measures to city finances (monitoring)
Local authorities are increasingly challenged to meet the dual objectives of stimulating economic growth while ensuring adequate provision of services to inhabitants. Monitoring city expenditures to determine their effectiveness in meeting these objectives can be a useful exercise.

Measuring the poverty impacts of programs and policies (evaluation)
By evaluating programs or policies (that may or may not be explicitly targeted at the poor) to determine their impact on the poor and urban poverty, city officials, planners and policy makers can learn and respond to the reasons for particular successes and/or failures.

Understanding specific urban issues
Often the objective of an urban poverty analysis will be to focus on a particular segment of urban life and services-such as land, crime and violence, environmental issues, slums, or urban transport. In other cases the emphasis may be on working with communities to identify their priorities and concerns.

Innovation
Using available data in creative ways, or finding unique approaches to collecting data may make urban poverty analysis more cost effective and may also uncover unique findings.

Integration of local partners
In addition to providing valuable information about a city, the process of conducting an urban poverty analysis can also offer opportunities for training and partnership building. Local partners may include local universities, NGOs, neighborhood groups and youth and women's groups.

Good overall mix
Many city poverty profiles are developed to address specific issues and target particular communities and neighborhoods. Conducting a comprehensive city-wide urban poverty analysis can be costly and time consuming and few cities have the resources available to do this.

Issues of local economic development
Some city poverty profiles may be designed and used to examine issues relating to local economic development-highlighting areas of slow or rapid growth, tracking infrastructure investments, and highlighting areas of potential development.

BOX 4: Lesotho: A Narrative Case Summary
In the Lesotho (1997) study, the objective of urban poverty analysis was to complement the existing and extensive work on poverty in Lesotho by illustrating (to policy makers) that urban poverty exists, and requires specific attention within national poverty alleviation strategies. Given that Lesotho is a predominantly rural contry, with only 20% of its population residing in urban areas, the majority of poverty assessments in recent years have largely ignored the unique features of urban poverty.
Where poverty is predominantly rural, there is a clear need to illustrate not only that urban poverty exists, but that it requires different types of policy interventions. To meet these study objectives, the researchers were careful in selecting both the type of data they collect and the persons from whom the data were collected. By income comparison alone, the urban poor are generally better off than the rural poor. However, to illustrate the multidimensionality and relative nature of poverty, the study combines income characteristics with non-income characteristics (access to services) and qualitative perceptions of poverty.
Within the broad category of urban poor, there are groups that have specific vulnerabilities that are not immediately apparent. To ensure that the stories of these vulnerable groups are told, the Lesotho study included a targeted sampling (most often by approaching respondents at their place of work) of seven preidentified vulneable subgroups: female headed households, street children, informal traders, piece job workers, shebeen men and women (part of the beer brewing industry), construction workers and their families, and retrenched mineworkers and their families. While most of the vulnerability of the targeted respondents can be attributed to insecure employment-conceptions of poverty were shown to vary by category of interviewees (i.e. being out of school is the main indicator for street children, homelessness is the most important indicator for piece job workers, and lack of basic household needs as the primary indicator identified by household heads.) Using a questionnaire (see Annex 3) with some open-ended questions, the Lesotho study measures poverty through 3 methods, (i) income indicators (comparing these with a World Bank poverty line); (ii) nonincome indicators (ranging from housing to community integration) and (iii) people's perception of poverty (interiew responses illustrated the relative understandings of poverty). Although the relatively small sample size used in the study does not allow for broad generalizations or comparisons, the mixture of targeted and untargeted qualitative and quantitative household surveys used in the study reveals some of the unique features of urban poverty: vulnerability due to commoditization, greater exposure to market conditions, erosion of community structures and greater 'individualization' of poverty, vulnerability as a result of crime, a preponderance of female-headed households and greater exposure to environmental risks . While urban poverty analysis is important in its own right as a tool to identify and advocate for the needs of the urban poor, it can be of even greater value if placed within a larger context. In the case of Lesotho, a summary analysis of national level forces illustrates that the logical push towards urbanization that would follow from high rural poverty is in fact being mitigated by the labor migration to South Africa (particularly mining). The 40% of rural households that do not receive remittances are also the 40% that are below the rual poverty line, illustrating that remittances play a central role in mitigating rural-urban migration. By highlighting the vulnerability implicit in relying on remittances to mitigate rural poverty, the study presents a compelling case for why attention to urban poverty and by extension, development is relevant in forming sustainable national poverty alleviation strategies.

Some Lessons and Recommendations
In this paper we have tried to pull together the various methods and approaches used to analyze urban poverty with an aim to provide some background and guidance to those interested in such work. The review is based on a number of existing urban poverty studies that have been carried out in a range of cities in developing countries. The emphasis was on the specifics of methodology and data rather than specific outcomes and analysis. The paper highlights key issues relating to urban poverty analysis to demonstrate how different approaches were used to address different issues (refer to Annex 1). Several general conclusions emerged from the review that would be useful to think about when designing a study of urban poverty as outlined below.

Design and Analysis
Defining and analyzing urban poverty can be complex as it has many dimensions. For this reason of complexity and by extension cost, most urban poverty studies will target only a particular aspect of urban poverty. When designing a study of urban poverty, it is useful to focus on what aspects specific to the urban poor that are of particular concern and they will affect the design and approach of the study. For example, to generate a poverty map or spatial poverty profile, a mapping of slums, or access to services, quantitative data collected through a census or household survey would be required. For focusing on issues of urban crime and violence and its link to poverty, supplementing quantitative data with a qualitative approach may be more appropriate. Several approaches may be combined to capture a more comprehensive analysis of urban poverty.
One of the central challenges of urban poverty analysis is the lack of available disaggregated data. Many national level surveys are only representative at the urban or possibly city level. This level of aggregation, however, does not tell us much about poverty within a given city. This means that those who wish to understand and address urban poverty issues often need to undertake their own studies to access the information required. Most of the studies we reviewed carried out some kind of data collection that is representative at a disaggregated level within cities, as well as representative for subgroups in the population. This allowed for spatial analysis, which is critical for urban planning and for targeting poor areas.
Once the approach has been defined, it is also important to ensure a clear link between the information collected and the application of that information in city policies and programs. The studies presented on Johannesburg and Bangalore have both been designed to support municipalities in using information and monitoring systems to perform their work more effectively. This objective has informed what type of data is collected, how it is collected and most importantly how it is analyzed and shared. Local capacity building has also been an objective in some studies (e.g. Johannesburg and Vientiane), and not only builds skills, but also contributes to future analysis and follow up in using the available information.

Data
When possible, it is advantageous to use existing data sources given the high cost of collecting data. This can include census data, facilities surveys, GIS, or over-sampling an ongoing household survey in urban areas. For example, in the study on Crime and Local Inequality in South Africa used existing census and household survey data to generate a poverty and inequality map, then linked this with existing crime data to focus on the linkage of crime and inequality. In Montevideo, existing household surveys were used to analyze increases in local level inequality over time. This was then supplemented with qualitative data to study aspects of social exclusion that were thought to be linked to problems of increased inequality at the neighborhood level. A majority of the studies carried out some kind of household survey to meet the specific needs of the study. Appropriate sampling is a challenge to ensure that the data are representative, and that the sample is large enough to capture the group of interest (e.g. specific urban areas, etc.). When designing surveys, it is useful to think about sustainability for follow-up surveys if monitoring progress in poverty reduction is an objective, as well as cost-efficient methods. In this paper, we have suggested that urban poverty analysis is most valuable when targeted toward specific issues. Similarly, such analysis is also most effective when conducted in partnership (international and local) with local universities and academics and with the intention of sustaining capacity in city governments to replicate the exercise.
Community level surveys can also provide a wealth of information for urban poverty analysis on services and facilities, particularly if the geographic unit is relatively small. Qualitative data can provide valuable local contextual information. For example, the qualitative poverty assessment in Vientiane illustrated the interdependence and informal cost-sharing for services between poor and non-poor households living in close proximity.
Once the data have been collected, it is important to make this information (both the data sets as well as the analysis and reports) publicly available. Given the high cost that goes into data collection and the huge value it can provide for research, any initiative should make provisions for making the data available on a widespread basis to researchers and others, such as the Atlas of Human Development project in Brazil (see Box 3).

Timing
An important consideration in designing and conducting and urban poverty analysis is the issue of time. There are two elements to consider. First, given that cities are dynamic and ever changing environments, it is important to understand how the duration of the study itself (from design through data collection, analysis, and reporting) will impact on the findings. For example if a study is launched shortly after civil unrest or a natural disaster, the findings that are published 1-2 years later may no longer be relevant to policymakers and planners.
Second, in designing and conducting surveys in urban areas, experience shows that respondents may have less time available for responding to questions and thus it is particularly important to keep interviews within a reasonable time frame. It is generally difficult to carry out a survey of more than one hour. In the Johannesburg Service Delivery study for example, the team had to review and revise their draft survey extensively to ensure that it could be completed in no more than 30 minutes. Time intensive surveys, while appearing comprehensive, may in fact produce poor quality data if respondents are not interested and annoyed by the process.

Cost
Rough cost estimates were available for a few of the studies carried out. The cost of data collection, both qualitative and quantitative, ranged from US$20,000 -US$60,000. This largely depended on sample size, length of the questionnaire, and local costs. Other costs would include staffing costs for technical expertise and data analysis, which can be significant if international experts are used. While it may appear difficult to justify the initial expense of conducting analytical work, the longer-term benefits of such an investment may prove very cost-effective. Urban poverty analysis may serve as the baseline of a project monitoring and evaluation program, facilitate the most appropriate targeting of activities and funds to achieve project outcomes, and provide valuable input into a city's information monitoring system.

Challenges in practice
Several cases illustrate the various challenges of carrying out urban poverty analysis and how these challenges were overcome. Many of the challenges were related to sampling, data collection, and definitional issues during survey implementation. Few, if any, cases have achieved a sustainable mechanism for collecting consistent information over time that can then be used for monitoring and evaluating policies and programs.
In Lesotho, for example, interviews were only conducted in selected low-income areas of three towns. Thus there was no city-wide data that could put the poverty analysis in the overall urban context. Within the low-income areas, however, households were randomly selected which did capture some differences in income levels. There was also concern, common to many other surveys, of a possible bias in sampling in that it is particularly difficult in urban areas (relative to rural) to find residents at home. Most surveys skip to another randomly selected household after 2-3 attempts to find someone at home. This may bias the results in capturing larger households (where it will be more likely to find someone at home), etc. If possible, follow up visits to the households should be scheduled at different times of the day. Finally, it is difficult to capture sub-groups in urban areas that may not have a fixed residence such as pavement dwellers, street children, informal traders, who are generally among the poor. Capturing these groups may require a specially designed instrument and approach.
In Ho Chi Minh City (1999) and other city studies, interviewers found that specific definitions were sometimes confusing. For example the definition of what constitutes a household or a kitchen, and description of physical structure are sometimes perceived differently by households and interviewers and thus need to be defined explicitly. There is also an emerging literature on the definition of slums and tenure, particularly in the context of the Millennium Development Goals (UN HABITAT, Cities Alliance).

Considerations for designing urban poverty analysis to meet project needs
From the review of cases of urban poverty analysis carried out in a range of circumstances using a range of methods, a suggested check list of issues to consider has emerged as follows: What is the objective of carrying out the urban poverty assessment?
What kind of information will be needed to meet the explicit objectives?
Is the information needed already available and/or can it be derived from existing data?
If the information is clearly not available, what resources are available for conducting an urban poverty assessment? (money, time, and staff) What types of instruments need to be developed and piloted for carrying the urban poverty assessment and for analyzing the findings? (surveys, questionnaires, data entry forms/computers, statistical software, etc.) What methodological approaches are most appropriate? Primarily quantitative data (Multi-topic or Single-topic Household Survey), Primarily qualitative data (Urban Participatory Assessment), or a Community Level survey ? Are there opportunities for mixing quantitative and qualitative approaches ?
Is sample data representative at a sufficiently disaggregated level to capture: intra-city poverty characteristics at the neighborhood, zone, municipality or district level; and subgroups in the population such as pavement dwellers and street children?
Who will carry out the data collection and are there mechanisms to ensure sustainability?
Are there issues to be raised in the urban poverty analysis that may be sensitive to community members, politicians, donors, etc? Is it possible to address these issues early on so that they will not threaten the outcome of the urban poverty assessment?
Is capacity building an objective? If so, what can be built into the effort to ensure that local partners receive as much training and participation as possible?
How will the analysis be linked to specific policy and projects?
What are the mechanisms for making data widely available for planners and researchers?

FOCUS Key objective of focus for data collection and analysis
Assess the level and status of urban poverty in Lesotho, place the findings of the assessment within the socio-economic realities facing Lesotho, and identify areas for intervention.

DURATION
October-April 1997 (6 weeks). NOTES: see Box 5. By building upon existing data, the study increases its relevance through filling identified gaps in knowledge. In order to get a 'well rounded' understanding of urban poverty it was useful to use a sampling mix that complemented geographic location with particular sub-groups.

FOCUS Key objective of focus for data collection and analysis
To test the feasibility of a rapid survey methodology and instrument to collect data on basic human development indicators. While the Lagos study was intended as preparation for national survey, the findings were utilized as a management tool for local authorities. CONTEXT: Pilot study for proposed National General Household Survey.

NOTES
The CWIQ approach allows for rapid collection and review of data, and may facilitate the ability of cities to collect and analyze data on a more frequent basis. This may be very useful as part of developing systems to monitor and evaluate programs and policies. However, this study also illustrates the challenges of identifying an appropriate sample size to make the data useful. In a city of 15 million, this presents an enormous challenge.

FOCUS Key objective of focus for data collection and analysis
To analyze how spatial polarization may be linked to 'social exclusion'. To identify the characteristics of poverty among vulnerable groups. CONTEXT: Conducted as part of a national poverty assessment of Uruguay. 6 neighborhoods selected by using recent census data to determine which areas had 'high risk' --low levels of human and social capital variables (i.e. educational attainment, socioeconomic composition of household, percentage of single teenage mothers).

COSTS and FINANCING:
Total costs for study; financing arrangements. $30,000 (qualitative data collection and analysis), $10,000 (quantitative data analysis (data set already available).

IMPLEMENTATION STRATEGY: Outline institutions responsible for implementation -local and international partners COMMISIONED and IMPLEMENTED by: The World Bank, and local consultants (NGOs) NOTES
This study highlights the need for disaggregated data for targeted policies. The findings illustrated that the poor in Montevido have become spatially polarized within the city and the effects of this polarization may not be apparent through aggregate data on income distribution. Uses a good mix of econometric and quantitative data to inform and build the case for the value-added of qualitative data. METHODOLOGY: PUA (participatory urban appraisal) which includes use of focus groups, seasonality analysis, trend analysis, causal impact analysis (diagrams), solar system analysis, and other participatory methodologies. PUA is described as 'an iterative, flexible research approach that is appropriate for investigation of complex causal relationships (such as those that surround violence in Jamaica)'

DATA and SOURCES: Type of data used; sources
To capture the relationship between poverty, social institutions and violence, the researchers framed their data collection around the notion of vulnerability as a function of 5 assets. DATA: Measurement of Assets: Labor, Human Capital, Productive Assets, Household relations, and Social Capital. SOURCES: Participatory Urban Appraisal.

SAMPLE SIZE, LOCATION and METHOD:
Specify sample size and method for determining sample. 5-communities within Kingston that exhibited different types of violence (determined by levels of drug activity, political affiliation, and economic circumstances.). Purposive sampling: identifies study communities considered representative of the issue under investigation-and uses in-depth investigation with a number of groups within the sample so as to be representative of the community.

COSTS and FINANCING: Total costs for study; financing arrangements.
Not Available

IMPLEMENTATION STRATEGY: Outline institutions responsible for implementation -local and international partners
COMMISIONED and IMPLEMENTED by: The World Bank,

NOTES
Violence in Jamaica is a predominantly urban problem that affects the urban poor disproportionately. This study offers an example of an urban poverty assessment that disaggregates with reference to a particular facet of urban poverty. Uses a highly participatory methodology to understand and analyze the complex causal relationships related to urban violence. Focuses on understanding the relationships between the 'assets' of the urban poor and the dynamic relationships between these assets and urban violence.
A clear finding of the report was the linkage between high levels of violence and lack of work (labor asset) and vice versa. Another finding illustrated the cyclical nature of 'area stigma' where high rates of violence make mobility within the community dangerous, thus resulting in reduced access to education (human capital) and lack of investment in communities (productive assets, social capital).

Brazil/ Rio de Janeiro: A City Study
World Bank. Rio de Janeiro, Brazil, 1999.

FOCUS Key objective of focus for data collection and analysis
To examine the municipalities' expenditures and programs in light of new challenges (declining economic development and increasing vulnerability of the poor) facing the city. Using this analysis the report provides suggested priorities and action areas to make Rio a more competitive, fiscally sound and pro-poor city. Three core areas of analysis: (i) Examine whether programs and expenditures are reaching the poor. (ii) Determining what lessons can be learned from other cities implementing similar programs and, (iii) Assessing the fiscal and institutional resources available to the city. CONTEXT: This report was written in collaboration with Rio City's Secretariat of Strategic Affairs with the express purposes of implementing the 1996 Plano de Metas (which was designed to operationalize an earlier participatory municipal strategy).

DURATION N/A INSTRUMENT and METHODOLOGY: Classify and describe instrument(s) used; methodology for implementation.
METHODOLOGY: This study is a comprehensive urban assessment, in which urban poverty is dealt with explicitly only in part. However the analysis of municipal programs and expenditures provides valuable insight into the quality and reach of services, and the examination of growth industries provides a useful discussion of job development as a poverty alleviation strategy. (Estimate) $30,000 of which approx. $15,000 for data analysis (mapping existing data with programs).

IMPLEMENTATION STRATEGY: Outline institutions responsible for implementation -local and international partners
COMMISSIONED BY: The World Bank. IMPLEMENTATION: In collaboration with Rio Municipality-but intended as Bank recommendations.

NOTES
Comparisons with other developing/developed cities to help analyze trends in data (for ex. industry to service ratios in economically dynamic cities) provide a useful and realistic grounding for setting objectives.
The data analysis illustrates that, contrary to popular and policy conception, the poor are NOT spatially concentrated in the favelas and that means-testing for effective targeting of social programs requires much better data on the characteristics and location of the urban poor in Rio.

Urban Livelihoods and Food and Nutrition Security in Greater Accra, Ghana
International Food Policy Research Institute et al., 2000 (research data from 1996, 1997, 1998).

FOCUS Key objective of focus for data collection and analysis
To provide an understanding of the nature of urban poverty and the relationship between urban poverty and food insecurity (or malnutrition) in Accra. This report looks at both food insecurity and child nutritional status-seeking to determine how the strategies employed by the urban poor to secure their livelihoods affect the household's food security; the care of household members, especially children; and the resulting health and nutritional outcomes. CONTEXT: The study report is a compilation of two related and published studies (Ga Mashie Study Team: 1996, Accra Study Team: 1998.

NOTES
This comprehensive study offers a very thorough examination of the complex causation relating to food insecurity-relating it to urban livelihoods and child-care. The study of the linkage between food security and child care is unique and required the development of an indicator to measure care and evaluate its impact on nutritional status in an urban environment.
A key finding from the report is that 'all the variations in income in Accra, as well as health and nutrition outcomes, is explained by factors that vary at the household and individual levels and not by community-level effects'. This has valuable implications for policy.

FOCUS Key objective of focus for data collection and analysis
To fill a perceived gap in knowledge about dimensions of urban poverty in Vientiane (national poverty assessment included only 2 households from Vientiane). As a qualitative assessment the objective of the study was to provide a 'description of poverty in terms of what and why rather than how much and how many. It seeks to describe the nature of poverty, the livelihoods of different types of poor people and the different experiences of poverty.' CONTEXT: Conducted as part of an analysis for the planning process of a Vientiane Urban Infrastructure and Services Project (ADB).

NOTES
Qualitative data is presented through a mix of charts and diagrams as well as narrative case studies--that present concepts such as social exclusion and vulnerability in a meaningful/tangible way. The data on respondent ranking of problems is disaggregated by gender and age illustrating the ways in which poverty is interpreted and felt differently by different segments of the population. In understanding the livelihood strategies of the poor, the assessment highlighted the interdependence between poor and non-poor households, offering valuable insight into service delivery policy, where targeting may fail to take into account of such relationships.

Poverty in Ho Chi Minh City: results of participatory poverty assessments in three districts (1999)
Save the Children (Tim Bond) Ho Chi Minh City, Vietnam. November 1999.

FOCUS Key objective of focus for data collection and analysis
To understand poverty and its causes from the perspective of poor people, and to understand their priorities and perceptions of poverty alleviation (both through their own coping mechanisms as well as through government programs). CONTEXT: Developed in part to complement the quantitative household survey data from the Vietnam Living Standards Survey (1997/1998). This study was one of four such studies carried out in Vietnam (coordinated by the World Bank).

INSTRUMENT and METHODOLOGY: Classify and describe instrument(s) used; methodology for implementation.
METHODOLOGY: This report is the compilation of a number of participatory poverty assessments that were conducted using the following methods:

NOTES:
As a participatory assessment the study emphasizes the voice of those interviewed, capturing these in illustrative and informative narrative story boxes. Focus group discussions on Anti-poverty programs encouraged participants to identify and rank the services that were aware of and used (in some cases respondents were unfamiliar with and thus not using available services). This can be a useful monitoring tool for those implementing such programs. The study highlighted neighborhood specific poverty issues and emphasized through evidence and recommendations the need for local authorities to have greater autonomy over implementing local anti-poverty programs.

Ho Chi Minh City Participatory Poverty Assessment (2003) Poverty Task Force/ Save the Children (Trin Ho Ha Nghi, Huynh Thi Ngoc Tuyet and Bill Todd) Ho
Chi Minh City, Vietnam. July-August 2003.

FOCUS Key objective of focus for data collection and analysis
To explore the grassroots perspectives on poverty trends and causes, and provide feedback on certain pro-poor policies initiated by the government, namely: (i) poverty reduction, (ii) grassroots democracy, (iii) public administration reform, and (iv) migration and urban issues.

NOTES:
In building on the reports of 1999 and 2001, this PPA identified the changing response by communities and local officials to the plight of poor migrants. Local officials recognize that poor migrants do not receive adequate access to social services, often pay considerably more (7-8 times) for basic services (water, electricity, etc.) and are among the most vulnerable. The study also illustrated how the influx of migrants were actually improving the conditions of non-migrant poor who were capitalizing on migrant need for accommodation and foodstuffs.
The study offers a unique lens on the Grassroots Democracy Decree by comparing community responses to the responses of officials on how the participatory democracy measures are being implemented. The findings reveal that there are still considerable gaps in the effectiveness of the policies. The review of the Public Administration Reform (PAR) highlights some of the practical challenges of PAR, particularly in the area of housing and land (where it is suggested that additional decentralization would be most efficient).

Cali, Colombia: Toward a City Development Strategy
The World Bank (Alexandra Ortiz) Cali, Colombia. 2002.

FOCUS Key objective of focus for data collection and analysis
To conduct a city-wide identification and prioritization of the city's main problems. To develop an analytical framework of the prioritized issues to feed into the city development strategy. While this paper provides insight into urban poverty in Cali (Ch. 4), it was not designed as an urban poverty assessment. CONTEXT: This report is intended as an information gathering advisory tool to the development of a World Bank funded, Caliimplemented City Development Strategy. . While the bulk of input into this report consisted of analysis of existing data, a Household Survey (1912 HH) was developed and conducted to assess satisfaction with municipal surveys (by EPSOC). This EPSOC survey will be described below.

SAMPLE SIZE, LOCATION and METHOD:
Specify sample size and method for determining sample. 1,912 households that were representative of 5 different geographic areas and 6 socioeconomic strata. The Colombian statistical institute defines socioeconomic status as a function of 'unsatisfied needs' (housing characteristics, etc) and is determined at the level of one square block of housing units. These strata are used for targeting subsidies. The survey gathered data on household income and included modules on Housing; Access to and satisfaction with basic services; Education and Health; the Labor market, Food security, Participation in city affairs, and Priorities of the population. COSTS and FINANCING: Total costs for study; financing arrangements.
(Estimated) $50,000 for data collection, plus an additional $50,000 for analysis of papers and compilation of final report.

IMPLEMENTATION STRATEGY: Outline institutions responsible for implementation -local and international partners
COMMISSIONED BY: The World Bank to support the city of Cali. IMPLEMENTATION: Studies commissioned from a range of partners, but analysis conducted by the World Bank.

NOTES
When the income data from the household surveys (EPSOC) was compared to the socioeconomic strata it was found that the strata did not provide an adequate measure of poverty (there was considerable income heterogeneity within all of the strata) and highlighted a need to reconsider the use of strata as a means to targeting subsidy programs. As part of the household survey, respondents were asked to rank their priorities with regards to municipal services. Comparing these preferences with municipal expenditures provided a useful tool for policy makers. Similarly, when municipal expenditures mapped by geographic areas were compared with information on geographic concentrations of poverty and wealth it was noted that a disproportionate amount of expenditures went to better off areas.

South Africa: Monitoring Service Delivery in Johannesburg
The World Bank (Vandana Chandra (Task Manager), Shashi Kolavalli and Bala Rajaratnam (Consultants), Johannesburg, South Africa. 2002 FOCUS Key objective of focus for data collection and analysis To pilot a simple and affordable survey that looks at how the status of service delivery changes on the ground, particularly in poorly served areas, and to integrate the generation and use of information with political processes to give voice to the city's poor residents. The long term outcome of this study would be the development and use (by service providers, policy makers and citizens) of a service delivery monitoring mechanism. CONTEXT: This study is part of a series of studies commissioned for the city of Johannesburg to look at issues of local economic development and service delivery. The City of Johannesburg has developed a number of strategic development plansthe most recent and far reaching being 'Vision 2030'.

DURATION
Between mid August and end September 2001.

INSTRUMENT and METHODOLOGY: Classify and describe instrument(s) used; methodology for implementation.
METHODOLOGY: Identify 'poorly served areas' or clusters (using 1996 Census data) with the same (poor) service delivery characteristics. Select survey sample. Develop questionnaire and conduct household survey using hand-held computers. INSTRUMENTS: The key innovation for this study was (1) cluster survey methodology for sampling; (2) the use of hand-held computers, or palm pilots that transferred data into Excel; (3) speed, affordability, accuracy and consistency in data analysis. The palm pilots made data collection easier (user-friendly interface), and more efficient (catching errors as they occur). Survey instruments were also integrated with GIS (Geographic Information Systems) data. A 30-minute survey was developed to collect information on access to basic services, household expenditure on services, household priorities, and educational/employment status of HH members. IMPLEMENTATION: 5100 households in 6 weeks by a team of 20 enumerators. DATA and SOURCES: Type of data used; sources DATA: Housing quality, Land, Toilets/sanitation, Water, Community priorities (regarding services), public infrastructure (telephones, waste removal, power supply, transport, health services, schools, police/ambulances/safety), Employment Status, and Payment for services/ Cost recovery. SOURCES: Multi-topic City Survey and Population Census (1996) SAMPLE SIZE, LOCATION and METHOD: Specify sample size and method for determining sample.
Cluster Survey methodology: After an initial assessment of the 1996 census data, it was determined that 921 enumerator areas, with 125, 409 households met the criteria of being 'poorly served' (in at least one of three services: water, sanitation and electricity). In the interest of keeping the survey to below 6000 households, only 20% of the poorly served EAs were surveyed, resulting in 5,100 household surveys carried out. Within the EA's, households were selected randomly and their ground location was identified on the GIS to enable production of detailed maps of how the status of service delivery on the ground changed between 1996 and 2001. COSTS and FINANCING: Total costs for study; financing arrangements.
Background work, methodology preparation, survey costs including sampling, programming of PALMs: $59,000 Total (including data analysis and report writing): $79,000

IMPLEMENTATION STRATEGY: Outline institutions responsible for implementation -local and international partners
COMMISSIONED BY: The World Bank IMPLEMENTED BY: The World Bank with survey and GIS consultants; financial support was also provided by the Municipal Government of the City of Johannesburg.

NOTES
In aiming to generate an 'information system' that will be directly linked into the service delivery and governance systems requires consideration of (i) how the data links (or can be linked) with existing and forthcoming measurements (census, etc.) (ii) what incentives are needed and obstacles faced in creating an information system that is useful, cost-effective and transparent.
The data collection instruments and the excel sheets that display the data were left with and shared among the relevant offices in the City of Johannesburg to facilitate next steps.

Action Research on Urban Poverty in Harare, Johannesburg and Luanda
The World Bank and CEFRE. Harare, Zimbabwe, Johannesburg, South Africa, and Luanda, Angola. 1999.

FOCUS Key objective of focus for data collection and analysis
To measure the level of urban poverty, social exclusion and social initiative in Harare, Johannesburg and Luanda, and establish criteria to improve future urban poverty programs and projects. The action research aimed to answer questions on (i) the extent and depth of social exclusion in the 3 cities, (ii) the relationship between social exclusion and urban poverty, (iii) the impact (and the extent thereof in the 3 cities) that civil society (social initiative) has in promoting urban social development, and (iv) the positive influence that policy can have in supporting social initiative (and combating social exclusion) as part of poverty prevention.

FOCUS Key objective of focus for data collection and analysis
This is a piece of analytical work that uses existing local crime data to determine 1. the theoretical underpinnings of crime (comparing sociological vs. economic models) 2. how the relative position of a community among neighboring areas may be associated with crime, and 3. whether crime is particularly prevalent in areas with high racial inequality.

DURATION N/A INSTRUMENT and METHODOLOGY: Classify and describe instrument(s) used; methodology for implementation.
METHODOLOGY: 1. Establish correlation between inequality and types of crime (across police jurisdictions) 2. Analyze data with respect to mean expenditure and income inequality (this correlates with sociological theories that suggest crime is a function of inequality) 3. Analyze data with respect to relative wealth between jurisdictions (to determine whether crime is influenced by the economic crime argument of cost-benefit) 4. Analyze data with respect to the relationship between crime and inequality with and between racial groups e) Conduct data analyses adjusting for under-reporting in crime DATA and SOURCES:

NOTES
The data collection exercise was framed by an understanding of why and how data can improve urban management (tax collection, service delivery, etc.). It also recognizes, however that for data to be put to use, training among relevant municipal staff is necessary.
By not limiting the survey to a narrow population group (e.g. a slum survey) or a single topic, such surveys can provide a baseline for urban welfare analysis at the sub-city level and across sectors.
The survey was designed so that the data (on at least household information and income and consumption modules) could be compared with past and future census data-adding to the collective data collection of the city and also so that it may be used in future studies to construct poverty maps.

NOTES
This work is unique in offering a comparative analysis of three cities using consistent data collection methodology in each of the cities.

Indicators and Data Sources
From: J. Hentschel & R. Seshagir. (2000), "The City Poverty Assessment Primer," Prepared for the Urban Management Course, Poverty Group, The World Bank.

Annex 3. Sample Questionnaire for Urban Poverty Assessment
The following questionnaire is from: Assessment of Urban Poverty in Lesotho United Nations Development Programme (UNDP). Maseru, Lesotho. December, 1997.