RIO DE JANEIRO, NOVEMBER 12-14
General Information
Wednesday, November 12
A. ORGANIZATION OF THE MEETING
The Sixth Meeting of the RIO GROUP on POVERTY STATISTICS was hosted by the Chair of the Group, the Brazilian Institute of Geography and Statistics (IBGE), in Rio de Janeiro, Brazil, during the period 12-14 November, 2003. The Economic Commission for Latin America and the Caribbean acted as secretariat of the Meeting. The agenda and the list of participants are included in Annexes 1 and 2.
B. OPENING SESSION
The President of IBGE, Mr. Eduardo Pereira Nunes, inaugurated the Meeting. He first highlighted the advanced stage of the work of the Group. Members have presented to the Meeting documents that summarize practices for the approaches to poverty measurement identified in previous meetings. Simultaneously, the government of President Luiz Inácio Lula da Silva, has given a central role in his policy agenda to the fight against poverty.
He also pointed out that IBGE is under a heavy demand pressure from the government, the Brazilian society, and the international community to produce the information needed to evaluate the magnitude and nature of the problem and to design, implement, and evaluate policies in this area. Numerous examples were given of demands of the Brazilian society to alleviate poverty and particularly extreme poverty.
In this framework, IBGE has been simultaneously a member and a user of the products of the Rio Group. Placed in the latter role it has been in a good condition to verify the usefulness of the production of the Group. Putting together practices developed under different approaches has been especially useful, due to the fact that such a multidimensional concept cannot be reduced to one figure or group of figures that correspond to one dimension of poverty. Different actors demand statistics that respond to their interests and views. The work of the Rio Group is therefore very satisfactory.
He went through the agenda analyzing its content and the stage of advance of work. He gave special attention to progress in the measurement of child poverty and the dimension of health. Within the challenges, he highlighted the one of international comparisons and statistical strategies to satisfy the increasing demand of information.
Finally he referred to the institutional relations to be established with the Project of the United Nations Statistical Division (UNSD). The Group and UNSD have been doing efforts to coordinate their work. He expected that during the meeting new operational steps would be identified, and that presentations to the 2004 session of the Statistical Commission would show a program of coordinated work.
C. PRACTICES FOR DIFFERENT APPROACHES
The Rio Group meeting, held in Rio de Janeiro between the 12th and 14th of November, aimed at compiling current practices in poverty measurement, which were synthetically presented by the Group's members.
Available syntheses of practices include absolute poverty lines, relative poverty lines, access to basic services and basic capital possession, relative poverty, subjective poverty, and a combined measurement of deprivation and low income. Each synthesis follows, as far as it was possible for the authors, the following scheme:
1. Introduction. Conceptual guidelines.
2. Standards and Resources
3. Availability of periodic calculations.
4. Technical characteristics: Similarity and differences among estimates
5. Challenges, options, and shortcomings
Previous to the meeting, it was agreed that the authors would concentrate in the aspects of Standards and Resources, Availability of regular established calculations, and Challenges, options, and shortcomings. The outline distributed to the authors as a guide for the preparation of papers is annexed.
The conceptual guidelines and their similarities and differences were discussed throughout the whole meeting, with a level of detail consistent with the availability of practices within each approach. There are also draft reports about the accumulated experience of the Group on three transversal topics: Poverty dynamics, international comparisons, and information strategies for poverty measurement.
1. Absolute poverty lines
Four different practices were introduced during the meeting: USA official measurement, the USA National Academy of Science proposal to replace the current measurement, Canada's Market Basket Measure (MBM), and ECLAC's practice.
In the subsequent discussion, ECLAC commented on practices of Latin American countries that closely followed its method. Although it was not the intention at this stage to discuss the conceptual aspects behind the measurement, some few remarks were introduced. Both official USA's and ECLAC's measurements are based on a subsistence food value estimate and on a complementary component that represents the value of other subsistence items. The USA NAS proposal adds to the food value item explicit values for clothing and shelter. Canada has two measurements that aim for a compromise between social inclusion and basic subsistence. The LICO's is a relative measure that places more emphasis on social inclusion and the MBM on subsistence. The latter represents the cost of a basket that includes a nutritious diet, clothing and footwear, shelter, transportation, and a complementary component that corresponds to other necessary goods and services.
The outline for the report distributed before the Meeting included the following potential items:
a. Food or nourishment.
b. Clothing.
c. Shelter.
d. Transportation.
e. Education.
f. Health.
g. Energy or heating.
h. Rest of expenditure after any or a group of previous items.
Among listed items, practices have been reported for food, clothing, shelter, and transportation. Energy or heating is considered in the estimate of shelter in the USA NAS proposal. The group has had no access to standards for education. The topic of health will be discussed separately due to the introduction in this Meeting of some concrete proposals for which estimates are available.
The geographical breakdown used in ECLAC estimates usually refers to at least metropolitan, urban and rural areas, though there are cases in which these breakdowns are not possible due to unavailability of information. There are also exceptions where more categories can be identified, among which the most distinguishable is Brazil, where poverty lines were estimated for 23 different regions or sub regions.
The Canadian MBM calculates 47 thresholds, based on province in size of areas. However, not every component of the MBM is available at that level. In many instances, prices are calculated in urban areas, but applied to rural components.
Official poverty lines in the U.S. are computed nationally and do not vary by location or urban/rural status. The NAS panel recommended that poverty lines should reflect differences in the cost of living throughout the U.S. In the absence of official U.S. inter-area comprehensive price indexes, the panel recommended that poverty thresholds should be adjusted for differences in the cost of rental housing across geographic areas of the country. They recommended that data from the U.S. decennial census, in which approximately one-sixth of U.S. households are asked a series of housing-related questions including rental expenses, would be a reasonable vehicle for developing geographic housing indexes that could then be applied to the housing portion of the poverty thresholds.
The updating of poverty lines is done in all 4 cases. The official USA poverty line estimates cover since 1959 up to now and therefore a time series is available. The variation of poverty thresholds over time follows the change in consumer prices, as measured by the CPI. The NAS panel recommended that poverty thresholds should be updated each year to reflect changes in expenditures for the basic good and services contained in the poverty budget. To smooth out year-to-year fluctuations and to lag the adjustment somewhat, they also recommended that the calculations should be based on the most recent 3-year average. Finally, for evaluation purposes, the panel also recommended, for the first few years of implementation, that alternative poverty thresholds should also be updated based on price changes. Census Bureau research into the effect of using the NAS measures has generally used both approaches for updating thresholds.
In the Canadian case the proposal is that when the basket will be final, the items of the basket of food, clothing and transportation should be priced annually. The rent component of the CPI will be applied between census years.
ECLAC's poverty line value is updated for periods not covered by an Income and Expenditure Survey employing changes in the Consumer Price Index (CPI) for food. Exceptionally, when changes of the CPI for food have been significantly lower than the general inflation, the cost of the non-food basket has been updated according to the variation in the CPI for non-food items; as a result, the coefficient that links food and non-food baskets is no longer fixed.
In all the 4 cases, the poverty line is compared with household income, a proxy of the resources of the household, to decide if the members of the household/family are to be considered to be poor.
Resources in the official USA estimates are defined as the total monetary income, before taxes, received by the family. Wages, self-employment income, transfer income (such as public assistance income), pensions (as well as disability and survivor payments), income from assets (interest, dividends, rent and royalties), educational assistance, child support payments, alimony, and financial assistance from outside of the household, as well as other forms of monetary income, are included in this definition. This definition does not include the effect of taxes or noncash benefits (such as subsidized rent or food stamps). Imputed rental income for homeowners, as well as the imputed value of free public goods and services are also not included. Thus, the definition of resources for poverty measurement used in the U.S. is not as comprehensive as the income concept recommended by the Canberra Group.
The NAS panel recommended a much more comprehensive definition of resources than the official U.S. poverty measure (total pre-tax monetary income). The NAS recommended a definition of resources that includes all sources of monetary income together with the value of noncash benefits (such as food stamps, free school lunches, and subsidized rent). The NAS panel also recommended that necessary expenses should be deducted from income. These expenses include: income and payroll taxes, child care and other work-related expenses, child support payments, and out-of-pocket medical costs.
In ECLAC's estimates, household income is defined according to the concept from National Accounts that mainly includes monetary incomes derived from current account transactions, including transfers in cash. Imputed income for owner-occupied dwellings is also included. An imputed value for free public goods and services has not been incorporated into the income concept. Expenditure level as an alternative approach for poverty measurement has not been used.
Canada uses disposable income defined as total income minus some non-discretionary expenses. Total income refers to income from all sources, including government transfers. Income taxes paid, social contributions (such as employment insurance or contributions to registered pension plans), child and spousal support payments, work related child care expenses, out of pocket medical expenditures are deducted from the total income to produce an MBM disposable income. Income is calculated at the economic family level (people related by marriage, blood or adoption.
The estimates based on absolute lines are within the most widespread in the measurement of poverty. The Group has gathered the experience of countries and institutions that have long time series using a similar methodology. Nonetheless the experience of other institutions and countries could prove valuable to give a broader scope to users.
The USA estimate and those of 14 Latin American countries are official estimates. The Canadian estimates are not official.
In the session of absolute poverty lines, these additional topics were discussed:
- It was stated that the nutritional requirements on which the absolute poverty lines are based should not be calculated for a population that is physically underdeveloped. Otherwise, there is a risk that the poverty line will be set too low, at a level that would not permit attaining the real potential of physical development. It was clarified that most absolute poverty lines in Latin America are based on international nutritional requirements, which are consistent with an appropriate nourishment and health status in developed countries. Nevertheless, in the application of these requirements, size and weight averages for the country have to be specified. To avoid using averages that are insufficient for a proper physical development, options such as considering only the people with the highest weight-to-size relation or adding a percentage to the global average were suggested. In any case, there was consensus that the nutritional standards should be updated, as the ones actually in use were estimated in the 1980's.
- Among other more specific comments related to the construction of absolute poverty lines under ECLAC's method, the alternatives for the selection of the "reference group" -whose consumption habits are reflected in the food basket- were discussed. In particular, the option of a national reference group was contrasted to the option of a reference group for each subnational context. In relation to this topic, there were comments about the appropriateness of adjusting incomes to reflect differences in price levels across subnational contexts, especially considering the lack of price information in rural areas.
- Recent estimates confirm that differences of poverty levels among regions of a country may be significant. The operational and methodological challenges are important but have been solved increasingly in the practices being considered. It was stated that when these calculations have become available, the political consequences of publishing them may be considerable, because they may force a redistribution of budgetary resources.
- Within difficult topics the equivalence scales constitute an important one where there is still no agreement within the Group. It was again suggested that the Compendium should present arguments and facilitate references without handling the topic in detail. Some of the debate is retaken in the section of international comparisons.
The Group, through PAHO, has now available documentation on estimates of standards and resources in health. Though not being systematic practices yet, they follow methodologies that are similar to that of the items of expenditure that have been introduced in previous paragraphs.
In standards they include estimates of health care needs and health insurances. They refer to initiatives being developed in the framework of the Canadian MBM, in the USA NAS recommendations, and in a joint effort of PAHO and the Peruvian INE. Topics under discussion include options to consider a mean of actual out-of pocket expenditure or a health insurance, and to estimate an imputed value of freely provided governmental health services. Sources of information being used and the geographical scope of estimates were presented. Preliminary figures for the estimates are available. In conclusion, basic methods exist and have been tested, and it should be decided how to include this work in the Compendium.
A table that compares similarities and differences between ECLAC, Canada's MBM and the USA NAS recommendation has been prepared and is included as ANNEX 3
2. Relative poverty lines
Three practices were introduced in the Meeting. The ones of Eurostat and ECLAC covered a large group of countries from Europe and Latin America, while the last one referred specifically to Canada.
The Eurostat indicators of poverty should be seen as part of a first set of 18 indicators that were adopted at the Laeken European Council in December 2001. They follow an official definition adopted by the Dublin European Council in 1984, which regards as poor: "…those persons, families and groups of persons whose resources (material, cultural and social) are so limited as to exclude them from the minimum acceptable way of life in the Member State to which they belong." This definition, whilst not operationally precise, clearly implies a multidimensional, dynamic and relative approach. In the case of poverty indicators a single representative value of central tendency, the median was selected as basis of a relative poverty line, within other reasons due to the fact of being the most stable of indicators of central tendency. Although wealth or consumption could have been selected, income is the variable used as a basis to define the threshold of poverty.
As mentioned the relative poverty line is used together with other Laeken indicators of monetary poverty such as measures to assess the duration of poverty (e.g. the persistent risk-of-poverty rate); the severity ('depth') of poverty (e.g. the risk-of-poverty gap); the distribution of income within the non-poor population (e.g. the Gini coefficient; the s80/s20 income quintile share ratio); and a new proposal for a specific indicator concerning the poverty risk of persons in work.
In the case of ECLAC the relative poverty line has been used more in the context of income distribution than in the area of poverty. It has been observed that the differences within Latin American countries according to the relative poverty line are remarkably smaller than in the case of absolute poverty lines.
The Canadian experience is based on the Low Income Cut-offs (LICOs) that have been produced since the late 1960s. LICOs have been employed by Statistics Canada for three decades to determine low-income prevalence for various socio-economic groups. They have been criticized because their methodology is difficult to understand intuitively and because they do not provide an appropriate base for inter-provincial comparisons, because they are calculated at the national level and do not properly adjust for provincial variations in the spatial distribution of the population.
The outline for the report distributed before the Meeting included the following potential items:
a. Percentage of the median household income or expenditure (per-capita or equivalence scales).
b. Percentage of the average household income or expenditure (per-capita or equivalence scales).
c. Budget standards as derived from items consumed by a majority of the households
d. Child poverty as a characteristic of inequality
Eurostat's standard is a risk-of-poverty threshold that corresponds to a percentage of the median income. In practice, it calculates and publishes poverty rates using several percentages (40, 50, 60, 70%) of the median and the mean.
ECLAC has used 50% of the median as the standard for its calculations, although for internal uses other percentages of the median and of the average income have been used.
The Canadian LICOs convey the income level at which a family may be in straitened circumstances because it has to spend 20% more of its income on food shelter and clothing than the average family of similar size. There are separate cut-offs for seven sizes of family - from unattached individuals to families of seven or more persons.
Summing up, the Eurostat and ECLAC measurements cover both median and average income calculations. Although the Canadian LICOs are not budget standards, they are related to this concept in the sense of establishing as minimum level of welfare according to what the majority has. Child poverty will be discussed separately in this report.
In this approach the information sources for standards and for resources coincide due to the fact that both pursue the measurement of household incomes.
Eurostat has paid special attention to the use of comparable estimates of income within countries. This could not be reflected in practice until 1994, when the pioneering 'European Community Household Panel' survey was launched, prior to which expenditure data from Household Budget Surveys was used. Alongside other variables, the ECHP collects information on net monetary income accruing to the household and its members from all sources - including work (employment and self-employment), private income from investment and property and social transfers received directly. No account is taken of indirect social transfers, loan interest payments, transfers to other households, imputed rent from owner-occupation and income-in-kind. This longitudinal survey was launched prior to adoption of the Canberra Manual and does not therefore allow full compliance: its successor, the EU-SILC, will permit greater consistency.
Once total household net income (or expenditure) is collected, the figures are typically converted to reflect differences in household size and composition, using an equivalence scale. Whilst the desirability of such adjustment is commonly accepted, various such scales exist and the choice is essentially arbitrary. They all assume a greater or lesser degree of sharing of household resources amongst household members. In the EU the so-called "modified-OECD" scale is used, which gives a weight of 1.0 to the first adult, 0.5 to any other persons aged 14 or over, and 0.3 to each child. The resulting value is attributed to each household member.
ECLAC's concept of household income is the same as the one previously described in the absolute poverty line section.
In the Canadian LICOs, income thresholds are calculated for two different income concepts: income before taxes and after transfers, and income after taxes and after transfers. The two series of thresholds have been published since the early 1980's, but SC recommends the use of the thresholds based on the income after taxes and after transfers for two reasons. First, this represents the full impact of the mechanisms of income redistribution in Canada: taxes and transfers. Secondly, purchases of necessities are done with "after tax-dollars".
There is no explicit equivalence scale in the LICO. However, the thresholds are calculated for various family sizes, which means that there is an implicit equivalence scale.
In the case of Eurostat, official estimates are available. The Laeken indicators and other statistics are compiled/produced by Eurostat from common sources (currently the ECHP, soon to be the EU-SILC). These are published by Eurostat in regular series (e.g. electronic database "New Cronos" and paper texts such as Eurostat Yearbook, Social Situation Report, IP&SE Detailed Tables, Statistics in Focus) and ad-hoc publications. They are also supplied to policy directorates within the Commission, for use in regular publications (e.g. Joint Inclusion Report, Joint Inclusion Memoranda for Acceding and Candidate Countries, Commission 'synthesis' report to the Spring European Council) and other purposes. For 15 countries series are available starting in 1994 and for 16 more shorter series are available.
ECLAC started estimates for 16 countries in 2000.
The Canadian estimate is not an official measurement of poverty and is available since the late sixties
The breakdowns of Eurostat estimates are multiple. The 'risk-of-poverty rate' indicator is established according to the analytical variables in the following list:
- Age
- Gender
- Household type
- Activity status
- Tenure status
This allows identification of the most vulnerable groups in society (e.g. the elderly (especially women); children aged 0-15; young adults aged 16-24; single person households; lone parents; families with more than 2 children; the unemployed; the inactive; tenants).
Eurostat already produces on a regular basis the following additional breakdowns of the risk of poverty:
- Main source of income
- Educational attainment level
Other breakdowns are also possible.
- Nationality (e.g. citizenship/ethnic origin/language
- Occupation (e.g. managerial/administrative/manual; e.g. permanent/ temporary)
- Health status
- Location (e.g. rural/urban)
ECLAC's estimates are only available at the national level and without other breakdowns.
The LICO estimate is available for five community sizes - from rural areas to urban areas with a population of more than 500,000.
Within the challenges of relative poverty measurement, one that outstands is the need to relate or complement the indicator with others that help to explain the specific circumstances of poverty. As will be commented later the need to combine monetary and non-monetary indicators is other way of expressing this challenge. Additional breakdowns, as in other poverty measurement approaches are a challenge where the specificity of the breakdown depends on the countries characteristics.
3. Access to basic services and basic capital possession
This approach is characterized by great disequilibria between proposals and practices. In fact most of practices are related to the so-called poverty maps, derived originally from the population and housing census. Due to this fact, the possibility to establish a frequent practice is limited to the frequency of this type of census. At the same time most of the indicators refer to characteristics of dwellings. Some methods that combine population and housing censuses with household surveys are available, making it possible to extend the types of indicators and increase the frequency of results.
In this meeting two practices were introduced. ECLAC presented a paper that follows the format of other practices while South Africa introduced its experience through a power point presentation.
The outline for the report sent to participants included the following proposed categorization for access to basic services:
a. Levels of education.
b. Levels and types of health services.
c. Safe water.
d. Sanitation facilities.
e. Unemployment benefits
Regarding basic capital possession, the outline included:
a. Dwellings minimum characteristics.
i. Quality of floor
ii. Quality of walls
iii. Quality of roof
iv. Availability or type of heating
v. Size. More than a number of persons per room
b. Neighborhood infrastructure.
i. Quality of urbanization
ii. Existence of facilities (schools, hospitals, public transport)
iii. Security
c. Land and water in the rural area. Extension and quality.
d. Human capital.
i. Level of education for economically active population
ii. Level of health. Nutrition, morbidities
e. Social capital.
Available practices gathered by the Group do not cover many of the proposed categories, even though all of them are expected to be correlated with poverty.
ECLAC paper refers to the so called Unmet Basic Needs (UBN) methodology, introduced in Latin America in the early eighties in order to take advantage of the population and housing censuses in the analysis of living conditions. It consists in evaluating if households have access to specific basic services or possess basic capital; therefore, it is a "direct method", "non-monetary method" which measures the actual satisfaction of needs rather than the possibility of satisfying them, as income-based measures do.
Dimensions considered under the standard version of the UBN methodology are related to the availability of information in censuses. Typically, the following basic needs are included: access to a decent dwelling, access to safe water and sanitation facilities, access to basic education and economic capacity to attain minimum consumption levels. The first and the latter are discussed under the item "possession of basic capital", while the others are treated as "access to basic services".
Each dimension under analysis requires defining at least one indicator and the threshold under which a necessity is considered unsatisfied. It has been common practice to select indicators that are significantly correlated with a general state of deprivation, proxied by very low incomes (to test correlation with low incomes, household surveys are employed). Therefore in many occasions they were used in Latin America to determine population or households under extreme poverty.
In the original version of the UBN method, results are summarized in a single index by adding households with at least one unmet basic, in a similar fashion to the headcount index for income poverty.
The standards used in Latin American countries for access to basic services were:
a. Levels of education.
Basic education is considered an essential requirement for the adequate integration of people into a productive and social life, and the attendance to an educational establishment has been chosen as the main UBN indicator for this need. Attendance does not automatically guarantee that the need of education is being fulfilled; quality of education and repetition rates, for example, should also be pondered. Nevertheless, information on these aspects is usually not available in censuses or household surveys. Other indicators may be employed, such as the educational level of adults or literacy rate
b. Safe water
Access to safe water in sufficient amounts is very important as it is closely related to the health condition of people. Access is determined not only by the availability of a water source in or near the dwelling but also by the means in which water is brought into the dwelling. When available, the distance from the water source to the dwelling is also taken into account
c. Sanitation facilities
To evaluate the adequateness of sanitation facilities two aspects have to be considered. One is the availability of a toilet and the other is the mechanism for evacuating residues. In choosing minimum thresholds for these needs, important differences may arise between urban and rural areas, as sewage systems are usually not available in the latter.
No indicators were available for levels and types of health services and unemployment benefits
For basic capital possession the indicators used were:
a. Dwellings minimum characteristics
Standard UBN method applications typically consider three dwelling's characteristics: type of dwelling, construction materials and number of people per room. The first one is usually a weak indicator, because categories contained in censuses and surveys tend to be too general. Regarding construction materials, separate categories are established for roofs, walls and floors of the dwelling. Number of people per room ("overcrowding") is also considered an indicator of the adequacy of a dwelling, and it is strongly linked to health status of the members of a household
b. Economic capacity
It is common to include in the UBN method an indicator of the economic ability of household members to satisfy their material needs. It takes into account the educational level of the head of the household together with the ratio of non-income earners to income earners in the household. This indicator tries to capture the opportunity of the head of the household to be employed and have an adequate income.
No indicators were available for neighborhood infrastructure, extension and quality of land and water in the rural area, human capital, and social capital.
Differences between urban and rural areas are taken into account when choosing minimum thresholds of satisfaction. For example, in the case of access to safe water, a source located more than 100 meters away from the dwelling may be considered satisfactory in rural areas but not in urban areas.
The resources available to satisfy basic needs are by definition those compared with the standard. The process of selecting a threshold is therefore simultaneous to the evaluation of the resources available in the household.
The greatest advantage of the approach is its ability to produce data for small geographical zones or specific population groups, thanks to the high representativeness of the source of information. In fact, it is common to find UBN maps (also called poverty maps) where the prevalence of each necessity is depicted independently.
On the other hand, this practice has several drawbacks. Among them, the most important are its low frequency, the narrowness of welfare dimensions captured by population census, and the difficulty of aggregating results into one poverty index.
4. Social Exclusion, Social Deprivation, Empowerment and other social and political dimensions associated to poverty
Some analytical categories such as exclusion, deprivation or empowerment, are usually considered as related to poverty, and in occasions are used as synonymous to poverty.
The available experience in the Group is mainly related to the exclusion from adequate income, access to the labor market and basic services. For these cases, the term "deprivation" seems to be operatively more useful.
The approach of access to basic services or basic capital possession discussed above is related to the one we treat here as deprivation. Nevertheless, there are some differences, which we will explore on the basis of the practice developed by the British Group associated to Professor Peter Townsend. The most important refers to the conceptualization of necessity, which in the case of the latter practice is essentially relative and defined by people through their opinions captured through surveys. Through a technical procedure described below, the dissatisfaction of those needs is assessed, in order to identify deprived or "poor" population.
The concept of lack of access to basic services and basic capital possession has sometimes been considered as conflicting with the use of income poverty lines. Contrary to that view, Dave Gordon recently presented a practice that combines both approaches to define the poor population. In what follows, we describe this method, which contains and goes beyond the use of separate deprivation indices.
The standard used to define monetary poverty, associated to the command of resources as in the absolute poverty line methods, is the level of disposable income. In the PSE survey command of resources was estimated using the standard UK National Statistical Office method - net usual total weekly household income (e.g. after the deduction of direct taxes).
Total income for an individual refers to income at the time of the interview, and is obtained by summing the components of earnings, benefits, pensions, dividends, interest and other regular payments. Gross weekly income of employees and those on benefits is calculated if interest and dividends are the only components missing. If the last pay packet/cheque was unusual, for example including a holiday pay in advance or a tax refund, the respondent is asked for her usual pay. No account is taken of whether a job is temporary or permanent. Payments made less than weekly are adjusted to obtain a weekly figure. Usual gross weekly household income is the sum of usual gross weekly income for all adults in the household. Those interviewed by proxy are also included (Bridgwood et al, 2000). Usual net weekly income is calculated by deducting direct taxes from the usual gross weekly income.
For equivalisation, Gordon, Pantazis and Townsend (2000) have proposed that equivalisation scales should be based upon budget standards results so that they are socially meaningful. In the Poverty and Social Exclusion Survey of Britain, the equivalisation scale shown below was used which is based upon the simplified relativities in the Low Cost but Acceptable (LCA) budgets for various 'idealized' household types (Bradshaw, 1993; Parker, 1998, 2000)
| Type of household member | Equivalence value |
| Head of household | 0.70 |
| Partner | 0.30 |
| Each additional adult (anyone over 16) | 0.45 |
| Add for first child | 0.35 |
| Add for each additional child | 0.30 |
| If head of household is a lone parent, add | 0.10 |
| If a household member is disabled, add | 0.30 |
The deprivation indicator approach aims to discover if there are people living below the minimum publicly accepted standard. It defines poverty from the viewpoint of the public's perception of minimum necessities, which no one should be without:
"This study tackles the questions 'how poor is too poor?' by identifying the minimum acceptable way of life for Britain in the 1980's. Those who have no choice but to fall below this minimum level can be said to be 'in poverty'. This concept is developed in terms of those who have an enforced lack of socially perceived necessities. This means that the 'necessities' of life are identified by public opinion and not by the views of experts or, on the other hand, the norms of behavior per se." (Mack and Lansley, 1985).
The approach is based on three steps. First, to identify what constitutes socially perceived necessities; second, to identify those who, because of a lack of economic resources, are forced to do without these necessities; and third, to discover at what levels of income people run a greater risk of not being able to afford them (the poverty threshold).
The first step was taken by building up a list of ordinary household goods and common activities. Respondents are asked "Please would you indicate… the living standards that all adults should have in Britain today. For each item which you think is necessary, which all adults should be able to afford and which they should not have to do without."
The second step was to ask people what items they already had or wanted but could not afford. Items defined as necessities by more than 50% of the population but which were lacked because of a shortage of money were then used to construct an initial deprivation index. The deprivation index was then refined using standard scientific methods to ensure that all the components were valid, reliable and additive (see below).
The third step, finding the poverty threshold, was taken by using multivariate methods to determine the income for each kind of household that maximized the differences between the 'poor' and 'not poor' and minimized the differences within the two groups ('poor' and 'not poor'). This is the 'objective' poverty line and households, which have to survive on this low level of income for any appreciable length of time, are highly likely to suffer from multiple deprivations (Gordon and Spicker, 1998).
We summarize here steps of calculation to illustrate the procedure
Step 1 - creating a 'politically' valid deprivation index
The deprivation index includes only those items that at least 50% of the population considered as 'necessities of life that everybody should be able to afford', thus obtaining 'political' validity (and face validity) for the deprivation index.
The list of potential necessities consulted gives an idea of their nature:
|
Beds and bedding for everyone
Heating to warm living areas
Damp free home
Visiting friends or family in hospital
Two meals a day
Medicines prescribed by doctor
Refrigerator
Fresh fruit and vegetables daily
A warm waterproof coat
Replace broken electrical goods
Visits to friends or family
Celebrations on special occasions
Money to keep home decorated
Friends or family round for a meal
Money to spend on self weekly
A television
A roast joint/vegetarian equivalent weekly
|
Visits to school e.g. sports day
Attending weddings, funerals
Meat, fish or vegetarian equiv
Insurance of contents of dwelling
A hobby or leisure activity
A washing machine
Collect children from school
Telephone
Appropriate clothes for job interviews
Deep freezer/fridge freezer
Carpets in living rooms and bedrooms
Regular savings for rainy days
Two pairs of all weather shoes
Presents for friends/family yearly
A holiday away from home
Replace worn out furniture
A dictionary
An outfit for social occasions
|
Step 2 - creating a preference free deprivation index
Items included in the deprivation index are only those that people 'don't have because they cant afford' them. This answers Piachaud's (1981) criticism of Townsend's Poverty in the UK index that the poor may chose to live in squalor rather than be forced to by a lack of resources.
Step 3 - creating a 'scientifically' valid deprivation index
In order to construct a valid deprivation index it is necessary to demonstrate that each of its components is a valid measure of deprivation. This can be complex; however, the fact that the majority of the population consider all of these items to be 'necessities of life' provides a-priori evidence for 'face validity'. The 'criterion validity' of the deprivation index can be demonstrated by ensuring that every individual component of the index exhibits statistically significant relative risk ratios with independent indicators known to correlate highly with poverty e.g.
- Ill Health (health in last 12 months was 'not good' and Limiting Long Term Illness)
- Subjective poverty measures ( income 'a lot below' the poverty line, income 'a lot below' the absolute and overall poverty line)
Step 4 - creating a reliable index of deprivation
After establishing that the individual deprivation index components are all 'preference-free', and 'politically' and 'scientifically' valid, it is necessary to establish that they also form a reliable scale. This can be accomplished through a classical test theory model by calculating Cronbach's Alpha (SPSS Reliability) for each deprivation item and removing all items in the index that would increase Alpha if the 'Item was deleted'
Step 5 - checking the revised index is additive
The components of any deprivation index should be additive e.g. a person or household with a deprivation score of three should be poorer than a person or household with a deprivation score of two. Some components of the index may not be additive, for example it is necessary to check that a respondent who 'can not afford' a hobby and a phone is poorer than a person who 'can not afford' a phone but has a hobby. There is no easy way to do this as the number of possible combinations with a 26 component index is huge (26 factorial) but it is possible to check that any two components are additive by looking at the second order interaction effects in a ANOVA with equivalised income as the dependent variable and all the components of the index as the 26 independent variables.
Finally a procedure is introduced to identify the combined poverty line that constitutes the limit between poor and non-poor.
The 'objective' combined poverty line can be defined as the division between the 'poor' group and the 'not poor' group that maximizes the between-group sum of squares and minimizes the within-group sum of squares. It can be identified using the General Linear Model (in one of its forms e.g. ANOVA, Discriminant Analysis or Logistic Regression), controlling for income, deprivation and household size and composition.
D. Transversal Topics
The group has gathered experience in transversal topics that are related to the practices presented in the previous section but that deal with other challenges: international comparisons, poverty dynamics and strategies for the improvement of information.
1. INTERNATIONAL COMPARISONS
International comparisons constitute an important aim of the international community and an unavoidable topic for a group that operates within the framework of the UN Statistical Commission.
Difficulties in comparing poverty measurements derive from the inexistence of standards for poverty measurement approved by the Commission, which are available for many other areas, the conceptual intricacies involved in comparing welfare levels, the presence of political considerations, and the relative shortage of experiences in this area.
This situation has been changing in recent years due to diverse events, such as the approval by the Copenhagen Summit of a definition of absolute and overall poverty, the establishment of quantitative goals for poverty alleviation in the Millennium Declaration and the adoption of concrete indicators to monitor the MDGs, and the fact that international organizations as the World Bank and regional organizations as EUROSTAT and ECLAC have been periodically publishing figures of poverty estimated in each case with similar standards.
The group has recognized the importance of including this topic in the Compendium, and several documents on this issue have been prepared. Nonetheless, there are important limitations that prevent the Group from arriving at conclusive solutions regarding international comparisons. Though there are areas in which active advances can be expected, as shown by the European experience and the Millennium Declaration, there are also topics for which any agreement seems far from the present situation. In any case, it is relatively clear that imposing a single worldwide assessment indicator is not currently feasible.
A basic requirement for performing empirical comparisons is that the concept of welfare captured by the sources of information is similar, at least in some extent. This can be a difficult task. As a reference, the Canberra Group has listed in its Report the items that should or could be considered in the measurement of income. It has also surveyed the availability of data for at least some of the most important income items in the national household surveys and other sources of information. If international comparability of income-based poverty measurements is to be attained, there needs to be a minimum agreement on the limits and content of the notion of income.
The most well known practice in the area are the "one PPP dollar a day" poverty ratios published by the World Bank. Other efforts at the world level have pointed toward the development of indicators that focus on human capabilities or deprivations in welfare dimensions such as nutrition, health, education and housing. A good example is the Human Development Index, contained in the Human Development Report, which combines data on longevity and education with income. Other international organizations produce indices for their areas of specialization that in principle are comparable at the world level. Nonetheless, these are usually averages for countries and they do not result in an index of deprivation. Examples of this type of indices include those produced by the International Fund For Agricultural Development (IFAD): the food security index (FSI), the integrated poverty index (IPI), the basic needs index (BNI), and the relative welfare index (RWI).
In the case of relative poverty, the poverty line calculated as a percentage of the median has become a widespread standard in Europe. Nonetheless, this practice has been criticized because the indicator, which looks at aspects of poverty and inequality at the same time, measures the income gap with the central trend, but not the trend itself.
The problem arises because this indicator is invariant with multiplicative transformation of income; that is, its level does not change when all individual income are multiplied by the same coefficient. Though this multiplicative neutrality assumption is essential in the measurement of inequality, it is in effect debatable in the realm of poverty: the fact that poverty can increase even when the income of every consumption unit raises seems to be paradoxical.
It has become increasingly accepted that the size and composition of the household should be taken into account in the measurement of poverty, either through the use of equivalence scales or household-specific budget standards. The fact that comparability needs homogenous concepts and procedures can sometimes translate into different choices of equivalence scale for each country. For example in Poland the expenditures increase proportionally with the family size and according to the few part of housing expenditures, which have been an evident character of "public good" inside household, these two things can suggest that Oxford scale (original OECD scale) is more appropriate to Poland than modify OECD scale comparatively to France.
Under the approach of absolute poverty lines, where standards are chosen to represent a minimum acceptable living condition, the selection of categories and items to include poses several problems.
There must be some common conditions that are important for the well being of the population of different countries at the same time. Deciding on a minimum set of common consumption items would allow to identify those households living in extremely disadvantaged situations. The selection of items should keep in mind that the consumption of one item could be restricted to a minority and at the same time it could be more common in another country.
The availability of sources of information is other central topic to make comparability possible. The present and potential availability of information will be dealt separately.
As mentioned in the section that refers to the absolute poverty approach, an effort has been carried out at the national level to define standards for items such as food shelter, and clothing. On the subject of food, the use of internationally accepted standards of minimum nutrition levels gives a basis to a comparable approach. To move towards a synthetic approach is a difficult task. Many have used the cost of buying a basic list of items as a standard expressed in a monetary value, the so-called poverty line or extreme poverty line. Another type of synthetic indicator used in some studies on developing countries defines the poverty threshold using the first component of an analysis of principal components.
Nonetheless the treatment of income deserves a special comment. Both relative and absolute poverty lines use income. The World Bank and others pursue using purchasing power parity as a basis for the calculation of the poverty line. Others prefer to establish both the standard and income resources in current prices. It is obvious that international comparability demands agreement on the unit of measurement.
Others are in favor of using budget standards, as an alternative to equivalence scales.
"It would be much more meaningful to produce low income statistics which show how many households do not have an adequate income to allow them to meet their basic needs (absolute poverty) and/or participate in the economic, social, cultural and political life of country in which they live (overall poverty). Low-income thresholds and statistics should measure adequacy and not be arbitrary thresholds and the most widely used method of achieving this goal is to use a Budget Standards approach.
A budget standard is a specified basket of goods and services which, when priced, can represent a particular standard of living. Budgets can be devised to represent any living standard (Bradshaw, 1993) and, for example, national statistical offices could produce budget standards, which correspond with the 'absolute' and 'overall' poverty definitions agreed at the World Social Summit by 117 governments. This would produce income poverty thresholds, which are both nationally and internationally meaningful.
Budget standards are probably the oldest scientific method of exploring low living standards. Pioneered by Rowntree (1901) in his famous studies of poverty in York, they have since been used in many countries to measure income poverty. For example, in the USA, at both national and state level (Orshansky, 1965; Watts, 1980; NYCC, 1982; Renwick, 1993; Citro and Michael, 1995) Canada (Social Planning Council, 1981) the Netherlands (Hagenaars and de Vos, 1988) New Zealand (Stephens, 1995) Hong Kong (MacPherson, 1994) Britain (Piachaud, 1979; Bradshaw, 1993; Parker, 1998, 2000) and Australia (Saunders, 1998).
The fact that these standards are calculated for different types of family compositions makes the use of equivalence scales unnecessary.
The proposals mentioned pursue making comparable measurements where income plays a central role. The Group has also received contributions that pursue international comparability in the area of the deprivation approach. They are also needed for international poverty comparisons and anti-poverty policy monitoring since poverty is not only dependent on personal/household income but also on the availability of public goods, e.g. clean water supplies, hospitals, schools, etc.
The proposals are based on a work carried out by the Townsend Centre for International Poverty Research in the UK, on behalf of UNICEF, which attempts to operationalise the absolute definition of poverty agreed at the World Summit on Social Development (see above) to measure child poverty in the developing world.
Comparable information on severe deprivation of basic human need amongst children is available from high quality micro-data from the Demographic and Health Surveys (DHS) carried out in 68 countries during the 1990s. The DHS are nationally representative household surveys with sample sizes of about 5,000 households and an estimated cost of $200 per household (Loup and Naudet, 2000). A major advantage of the DHS is their random cluster sampling methodology. On average, 3,000 to 9,000 women of childbearing age were interviewed Deprivation can be conceptualized as a continuum which ranges from no deprivation, through mild, moderate and severe deprivation to extreme deprivation at the end of the scale. Figure 2 illustrates this concept.
Figure 2: Continuum of Deprivation
In order to measure absolute poverty amongst children using the World Social Summit definition, it is necessary to define the threshold measures of severe deprivation of basic human need for:
- food
- safe drinking water
- sanitation facilities
- health
- shelter
- education
- information
- access to services
Comparable information on severe deprivation of basic human need amongst children is available from high quality micro-data from the Demographic and Health Surveys (DHS) carried out in 68 countries during the 1990s. The DHS are nationally representative household surveys with sample sizes of about 5,000 households and an estimated cost of in each country (average 5,400) and each survey contains between 150-300 clusters, with an average of 200 clusters. Cluster size is around 2-3 km or smaller in urban areas (Gerland, 1996). Figure 3 demonstrates the wide coverage of Demographic and Health
Figure 3: Distribution of Demographic and Health Surveys
Note: Chinese data is from the China Health and
Nutrition Survey (http://www.cpc.unc.edu/projects/china/)
Table 2 shows the operational definitions of deprivation for the eight criteria in the World Summit definition of absolute poverty that have been used for the UNICEF study of child poverty using DHS microdata.
| Table 2: Operational definitions of deprivation for children |
| Deprivation | Mild | Moderate | Severe | Extreme |
| Food | Bland diet of poor nutritional value | Going hungry on occasion | Malnutrition | Starvation |
| Safe drinking water | Not having enough water on occasion due to lack of sufficient money | No access to water in dwelling but communal piped water available within 200 meters of dwelling or less than 15 minutes walk away | Long walk to water source (more than 200 meters or longer than 15 minutes). Unsafe drinking water (e.g. open water) | No access to water |
| Sanitation facilities | Having to share facilities with another household | Sanitation facilities outside dwelling | No sanitation facilities in or near dwelling | No access to sanitation facilities |
| Health | Occasional lack of access to medical care due to insufficient money | Inadequate medical care | No immunization against diseases. Only limited nonprofessional medical care available when sick | No medical care |
| Shelter | Dwelling in poor repair. More than 1 person per room | Few facilities in dwelling lack of heating, structural problems.
More than 3 people per room | No facilities in house, non-permanent structure, no privacy, no flooring, just one or two rooms.
More than 5 persons per room | Roofless - no shelter |
| Education | Inadequate teaching due to lack of resources | Unable to attend secondary but can attend primary education | Child is 7 or older and has received no primary or secondary education | Prevented from learning due to persecution and prejudice |
| Information | Can't afford newspapers or books | No television but can afford a radio | No access to radio, television or books or newspapers | Prevented from gaining access to information by government, etc. |
| Basic Social Services | Health and education facilities available but occasionally of low standard | Inadequate Health and education facilities near by (e.g. less than 1 hour travel) | Limited health and education facilities more than 1 hours travel away | No access to health or education facilities |
Children who suffer from severe deprivation of basic human need as shown in column 4 of Table 2 are living in absolute poverty as defined at the World Social Summit, i.e. Absolute poverty is a condition characterized by severe deprivation of basic human needs, including food, safe drinking water, sanitation facilities, health, shelter, education and information. It depends not only on income but also on access to social services."
A very important strategic topic is to decide how to move towards international comparison. At least two important decisions should be adopted.
The first is to start with a regional or sub regional comparison, as has been the case of the European Union and ECLAC, or to start since the beginning with a worldwide approach.
The second is to favor an international statistical instrument for information gathering, such as a special survey, or to favor transformations of available instruments.
There are no definitive positions of the Group as a whole in these matters, which still require further analysis.
2. POVERTY DYNAMICS
The study of poverty dynamics is crucial for understanding the origin of poverty and attempting to reduce it. Every approach to measuring poverty would benefit from a follow up in time of the units that at a certain moment were declared poor. Due to this fact, the Group has considered poverty dynamics as a transversal topic.
The sources of information have been an important obstacle to follow the poverty situation of households and individuals through time. Cross-sectional surveys, which are the most common in developing countries, allow the study of poverty dynamics, but only in a limited way. In this case, time series usually refer to the total number of poor people, but not to the number of people who became poor or escaped from poverty. These and a wide variety of other types of analyses are possible when longitudinal surveys are available. An intermediate option is viable in certain occasions, when in non-longitudinal surveys the sample is replaced gradually, allowing a partial follow up of households for limited periods of time.
On this topic, the Group has gathered experiences mainly from developed countries. Most of them refer to European countries and Canada, and are based on a common longitudinal survey in Europe and on an especial survey of Statistics Canada. Several important topics have been studied with these surveys. One of them is the persistence of households in poverty, for which indicators and standards have been included in recent EUROSTAT publications. Among the factors associated to the level and persistence of poverty, the following have been examined: geographical location, size of cities, type of households, head of household's level of education, average number of occupied persons, and age of occupied persons. Explorations of the association between persistence in poverty and the notion of permanent income, and of the relation between public unemployment policies and working poor, have also been performed, and are topics on which the Group has accumulated valuable experiences.
Some of the available studies intend to identify and analyze certain groups with especial characteristics within the poor. Such are the cases of working poor -for which a document that considers international comparisons is available- and recent immigrants in Canada. Furthermore, in occasions the studies are oriented towards judging the effect of public policies in relation to poverty. The fact that longitudinal surveys allow to make a follow up of individuals through time is relevant to study the effects of education, public policies and other factors on the performance in terms of employment, salaries, and access to goods and services. They are also a useful tool for analyzing policies associated to equality of opportunities, as well as for examining the relation between characteristics of the household and children, and therefore the chances for the reproduction of poverty.
There are certain causes for important changes in the situation of individuals, households and social groups that are not always captured by the most frequent longitudinal surveys. They refer to shocks originated in macro or micro economic changes or in modifications of the health, working or civil status, or other personal conditions of individuals. In the eighties and nineties, important structural changes have occurred in the economic and social areas, and it is necessary to improve the methods through which they have been measured.
Other areas of research in which Statistics Canada has been working are:
- When a person (or family) falls into poverty or rises out of poverty, how often is it related to labour market changes and how often is it related to changes in family composition? At the level of the individual, changes in family composition are more important than changes in jobs held by parents. However, changes in family status are relatively infrequent compared to labour market changes. Parents are much more likely to lose or find jobs, and experience changes in hours worked or wages, than they are to marry or divorce. When this is accounted for, in the aggregate, flows of children into and out of low income are associated roughly equally with family compositional changes and changes in wages and hours worked.
- When a man dies, how well does his wife recover from an income perspective? Public social insurance schemes are very helpful for older widows. When one includes all sources of income, income-replacement rates are much higher than one would predict based on benefits from public programs.
In the case of the U.S., the Census Bureau has produced a series of reports from the Survey of Income and Program Participation (SIPP) that have examined issues related to the dynamics of poverty in the country. The most recent report looked at the dynamics of poverty over a 4-year period of 1996-1999 (see http://www.census.gov/hhes/www/sipp96/sipp96.html for more details). The report examined issues such as:
- characteristics associated with transitions into and out of poverty
- median length of poverty spells and how poverty spell length differs by characteristic
- percent of the U.S. population, by characteristic, that experience poverty based on a monthly income measure, an annual one, and a
- "chronic" definition of poverty (those who are poor for every month of a 4-year period)
3. STRATEGIES FOR THE IMPROVEMENT OF INFORMATION
The statistical offices and other institutions that belong to National Statistical Systems are today under an increasing demand for poverty measurements. Most of the original measurements were done using an existing source of information. From there on, the growing demand derived in the need to include new questions in existing surveys, adding new surveys and increasingly recurring to administrative sources and National Accounts. Many countries have been obliged to review their information system in order to adapt it to the new requirements. Moreover, recent international events commented previously make the Group foresee an increasing need for information strategies that help to deal with the growing demand.
The group has gathered information on different strategies, of which the one from the European Union is probably the most structured. Two other participants have presented their experiences on information strategies.
The first is ECLAC's experience in promoting the expansion and enrichment of information sources in Latin America. The provision of technical assistance and the organization of regional discussion workshops and training courses have been the major channels of cooperation of the institution, to a great extent through the work of the MECOVI-Program, co-sponsored by the World Bank and the Inter-American Development Bank. Among other accomplishments, a significant increase in the coordination of the different surveys has been achieved in some cases. Simultaneously, international institutions are now working more systematically with the National Statistical Offices, instead of establishing joint ventures with other institutions.
The second experience presented is the Australian Institute of Health and Welfare's use of administrative information derived from the social programs that focus on people covered by governmental safety nets. The AIHW undertakes detailed analyses of official surveys (conducted by ABS), conducts a small number of direct collections, and manages and examines administrative by-product data sets (e.g. hospitals, disability, aged care, housing). Their role in terms of administrative information is important because they have close contact with potential users of public policy. They are able to produce data that assists policy makers and program administrators assess the effectiveness and efficiency of government programs and policies. These data sets can include the description of clients/users of government programs, where they come from, what they need, what they are provided with, where they go to, the extent of service use, clients' use of services across government programs, and the unmet needs. Therefore this information strategy has the potential to collect data on who is poor or socially excluded and why, as well as measure the impacts on people of one or a number of programs designed to alleviate poverty.
Apart from helping to establish information agreements with government departments administering social programs, the AIHW's main efforts are directed to define concepts, generate classifications of community services and gather the information. More recent work has seen the AIHW play an important role in data integration using common administrative items across public program statistical collections and through the development of linkage keys based on names and demographic characteristics.
Among the benefits of this kind of information are not only its low cost, but also its direct applicability to the evaluation of the social programs. In addition it can allow for comparisons between different service delivery strategies within or across social programs and the measurement of of service use within and across social program by clients over time. For example the data collection on major social program response in Australia to homelessness has been able to provide information on the causes of homelessness as well as how successful the program has been at ensuring clients of the program don't become homeless again.
It was shown that this strategy requires a significant amount of of organisation and goodwill between a number of parties before it can be effected. A considerable effort to develop standards, the support of service providers, the development of collection systems and the observance of important ethical considerations are also necessary. The experience shows that the use of administrative data is increasing in Australia, and that the implementation of data linkages is gaining momentum. It was also pointed out that the strategy to unify information production and policy implementation requires having social programs in place, condition that reduces its applicability in developing countries.
In the Fifth Meeting of the Rio Group, the ABS introduced a proposal for designing and offering countries an international standard survey for the measurement of living conditions, as an instrument for producing worldwide comparable information. It should complement other available sources of information such as housing and population censuses and employment household surveys. As stated in the proposal, an instrument of this kind would require a clear definition of indicators, standards, sources of information, etc. and of their link to measurement methods, and that constitutes, as in the case of the other initiatives, a challenge to be resolved. It was explained that this application should be part of an international effort of cooperation with institutions currently working in this type of activities.
During the discussions surfaced the importance of taking into consideration the experience of poor countries in the production of information. Some of the countries and institutions that are members of the Rio Group have advanced statistical systems or sizeable resources to improve them. For example, even though not every country in the European Union has access to administrative data, EUROSTAT is able to provide funding for the development of this information source in particular cases. Therefore, great care must be taken in order to produce a Compendium that also includes practices that do not require such abundance of information.
Finally an important initiative to produce a handbook on poverty statistics was presented to the group The representative of the UN Statistics Division (UNSD) from New York made a detailed presentation on a project to prepare a publication on poverty measurement. She referred to the UNSD website for more detailed information.
http://unstats.un.org/unsd/methods/poverty/default.htm
The representative of UNSD emphasize the project would build on the extensive useful material which had been generated by past Rio Group meetings. She invited the Rio Group to enter into a dialogue with UNSD to reflect together on the most effective way of future cooperation. She also sought support from the Rio Group for a planned UNSD workshop on poverty statistics to be held in the first half of 2004 in Latin America. The forthcoming session of the Statistical Commission will offer an appropriate forum for further discussions also with a larger group of stakeholders.
1 See http://www.measuredhs.com/
2 See http://www.measuredhs.com/
