GISdevelopment.net --> Proceedings --> GISDECO --> 2004
Abstract
Policy makers are currently encouraged to introduce area-based policies to target deprived areas, then set priorities, and reallocate resources. GIS-based indicators can be a valuable tool to describe differences in the quality of life and access to services and identify trends. GIS-based indicators provide valuable information especially if they are policy related, and to some extent they are able to generate better decisions and evaluate the policy performance. This paper explains how the integration of GIS into policymaking can be achieved using intra-urban indicators and shows through a case study the advantages of GIS in constructing them. GIS-based indicators are constructed combining different data sources such as census and administrative data. This paper also analyses how policy makers perceive the problem of urban inequalities and how they deal with data and indicators on decision-related issues.
To succeed in the GIS adoption and capacity building in developing countries, GIS should be able to respond to the local needs and it should be policy demand driven and application context sensitive. The use of intra-urban indicators can contribute to the integration of GIS into policymaking as well as meet local problems and needs.
Introduction
By means of a case study this paper explains how the integration of GIS into policymaking can be achieved using intra-urban indicators. This section introduces the problem of inequality within cities and the growing demand of information to target disadvantaged areas. The second section describes what indicators are, their functions in policy making and why GIS can help in its operationalization. The third section deals with the case study area and its planning context where the methodology to describe the inequality phenomena with GIS-based indicators is applied. The fourth section starts by analysing the inequality aspects that policy makers consider more relevant and how they deal with indicators. Finally, it is shown the operationalization of a set of indicators and how they can help to identify gaps and rank areas.
The “Global Urban Observatory”, the UN-HABITAT international capacity building network, indicates that many cities suffer from an information crisis that undermines their capacity to develop effective urban policy. It also warns that these cities do not have a sustained or systematic approach to assess the urban problems and cannot evaluate the success of the implemented policies. Urban indicators are seen as a tool that can improve that situation (Moor, 2000). There is also a recognition that GIS can be used for the collection and analysis of urban indicators. A proof of that appreciation is that in February 2003 ESRI donated 15 million dollars to the Global Urban Observatory’s indicators programme .
Since the Agenda 21 declaration, the importance for a sustainable development of reducing inequalities and disparities within cities it is constantly mentioned (United Nations, 1992; European Commission, DG XI (1994) in Mega, 1996; UNCHS, 2001). Spatial inequality occurs in urban areas around the world. However, especially in cities in developing countries, inequalities in the habitat conditions or access to social and physical infrastructure are particularly evident. At the same time, local governments are encouraged to target those deprived areas. Problems related with deprivation or poverty show spatial concentrations in cities accentuating the problems suffered by people living in certain areas (Pacione, 2001 p. 291).
Socio-territorial indicators are being used since the beginning of the twentieth century. In the 1970s, there was a growth in studies of patterns of inequality and spatial injustice with an interest to influence public policy. Later on, postmodernism emphasized or “celebrated” diversity and difference (Smith, 1994). In the 1990s there was again a growing engagement of geographers with inequalities, moral and social issues, including the theme of the ethics of professional practice (e.g. Couclelis, 1999). This growing concern is particularly evident and it is reflected in different reports and initiatives of international organizations that are stressing the importance of monitoring spatial inequalities within cities (e.g. UNCHS, 1995; The World Bank, 1996; EC (European Communities), 2000; European Commission, 2000; UNDP, 2000; UNCHS, 2001; UNDP, 2001).
Even though there is growing literature about inequalities and attention for the importance of measuring it; there is not enough discussion about methods and tools (basically indicators), how suitable they are, and the use of GIS to operationalize them. This concern is also expressed as a problem of lacking of “spatially relevant indicators” (Kunzmann, 1998). Several studies of patterns of inequality and spatial differentiation within the city have been based on indicators of personal income (e.g. Smith, 1973; Smith, 1994; Chakravorty, 1996). However, as Knox and Pinch (2000 p.100) put it “While socioeconomic differentiation is arguably the most important cleavage within contemporary cities, it is by no means the only one”. Besides, for most cities there is also a data availability restriction, since data of personal income is not collected at low levels of aggregation. This is the case of many cities in the developing world as well as in developed countries (see: Broadway and Jesty, 1998; Wessel, 2000).
Some of the criticism to “area-based” policy strategies to reduce inequalities argues that targeting is not based on needs and that the areas are not homogeneous (the ecological fallacy problem). To minimize the problems (and criticism) of ecological misleading outcomes, this research focuses on the use of small area units and low levels of aggregation and places stress on “need” more than any other distributive criteria. Furthermore, the construction of indicators outside a policy perspective is also widely criticised (Innes, 1990). Consequently, this research proposes to link the development and use of GIS-based indicators with policy.
The city of Rosario (Argentina) has different levels of housing needs and access to physical and social infrastructure. Rosario is taken as a case study and GIS-based indicators were used to measure spatial inequalities. Census data from 2001 was analysed together with
administrative data from the Public Housing Office. GIS-based indicators provide valuable information particularly if they are policy related and able to generate better decisions. For that reason, through a series of interviews it was analysed how policy makers perceive the problem of urban inequalities and how they deal with data and indicators on decision-related issues.
Overcrowding, education level, employment, tap water inside the house and access to primary schools were the five most important aspects of inequality according to policy makers. Diagnose and problems identification was the phase within the policy cycle where they would more appreciate the use of indicators. After the analysis, it was possible to identify the most disadvantaged neighbourhood areas and to measure the gap between the best off and worst off areas. Needs derived from indicators constructed from census data were complemented with point maps showing directly expressed need which proves to show “hidden” needs not captured by the derived indicators. Finally, factor analysis was applied to define a socio-economic factor which can be used to describe the socio-economic gaps in the city and as a proxy indicator of income.
Policy makers are currently encouraged to introduce area-based policies to target deprived areas, then set priorities, and reorient resources. GIS-based indicators prove to be a valuable tool to describe differences in the quality of life and access to services. To succeed in the GIS adoption and capacity building in developing countries, GIS should be able to respond to the local needs and be policy demand driven. The use of intra-urban indicators can contribute to the integration of GIS into policymaking as well as meet local problems and needs.
What are indicators and why GIS can help in its operationalization?
Indicators simplify complex phenomena into quantifiable measures that can be used for policy and decision-making. Therefore, indicators have three functions: to simplify, to quantify, and to communicate (Van Delft, 1997). Innes (1990) considers that an indicator focuses and renders intentionally selected areas of the reality. She puts it in this way: “An indicator is simply a set of rules for gathering and organizing data so they can be assigned meaning”
From these definitions, we can see the potential that GIS has to operationalize indicators. While constructing indicators, it is necessary to: organize data, to quantify and to communicate. These three indicators functions coincide with the usually acknowledged advantages of GIS: data organisation, spatial analysis and visualisation (Burrough, 1986 in: Huxhold, 1991; Webster, 1993; Ghose and Huxhold, 2002). Furthermore the spatial dimension of urban inequalities and the area-based policies that target deprived areas makes decision support suitable for analysis and monitoring with the use of GIS based indicators. Besides, in the late 1990s the value and potential of GIS to construct intra-urban indicators is increased by a combination of: a growing concern on intra-urban inequalities, the implementation of area-based policies and the developments in ICT and GIS technology.
Apart from rendering intentionally selected areas of the reality (Innes, 1990) GIS-based indicators can perform different functions which are related to the different phases in the policy cycle. Indicators can be classified as follows :
GIS-based indicators to measure spatial inequality
GIS can be used to operationalize indicators in any of the policy cycles. Besides, it should be established which policy goal is considered and which phenomenon the indicators will describe. In this case study, indicators are used to describe spatial inequality and can be related to the policy objective of targeting intra-urban inequalities.
In this research and to justify that inequalities in cities really matter it is necessary to consider an approach from a social justice perspective. When there is a need to monitor or describe inequalities, it is implied that with the use of planning tools there is the intention to change, improve and/or solve this problem. Hence, there is a concern with what the city should be, and some form of redistributive or compensatory action. According to Smith (1994 p.2) “questions of social justice, morality or ethics are usually described as normative, concerned with what should be, as opposed to positive knowledge which is about what actually is”. Smith (1994) considers that justice involves treating people fairly, which in distributive justice means that whatever is being distributed should go to people in the right quantities. He expresses that “fairness” means that people in the same circumstances should be treated in the same way. A difference can be made between arithmetic equality and proportional equality. In arithmetic equality everyone is getting exactly the same quantity of something and in proportional equality the distribution is justified according to a certain criteria such as need or market demand. In this research a need approach within a social justice perspective is taken.
The spatial patterning of inequality and difference results in the segregation of certain segments of the population (Langlois and Kitchen, 2001). Actions against that problem are usually addressed with area based policies that target those deprived or segregated areas. Social justice it is concerned with the question of who gets what where and how, and more precisely who should get what where and how (Smith 1977 in: Pacione, 2001). We can clearly see from these definitions that the spatial aspect of inequalities justifies the use of GIS-based indicators.
In this research, spatial inequality is considered to be a heterogeneous, multidimensional and complex phenomenon with several aspects. Two axes are distinguished with the following domains:
(un)equal conditions of quality of life
Criteria for constructing GIS-based indicators
After establishing the domains that cover the inequality phenomena, it is necessary to select a particular set of GIS-based indicators. The selection is based both on the selected domains and on effectiveness and efficiency criteria. The following UN-HABITAT criteria (UNCHS, 1995 p.18) were considered:
Following these criteria and both to relate the selection of indicators to local policy, and to integrate GIS into policymaking, in this research the selection of indicators was derived from present policy interests in relation to inequality. For that reason, interviews with policy makers were carried out. To be able to construct geographically disaggregated indicators two data sources were selected: census data and administrative data. The study area for the application of the methodology is the city of Rosario (Argentina).
Case study area characteristics and planning policy context, Rosario (Argentina)
Rosario is the third largest city in Argentina and it is located along the shores of the Paraná River at approximately 300 km northwest of Buenos Aires.
Rosario is the main city within a metropolitan area which has a population of around 1,300,000 inhabitants (“Gran Rosario”). The city of Rosario itself, according to the 2001 census, has a population of 923,424 persons living in an area of nearly 180 km2. In 1970s a crisis affected the economic structure of the region (Plan Estratégico Rosario (PER), 1998). As in many other cities around the world, Rosario suffered the effects of deindustrialization. The adjustment and reconversion of some industries (e.g. iron, paper, steel and chemical industries) “put the region in a critical economic and social situation, with difficulties to compete in a frame of open economy context and international competition” (Plan Estratégico Rosario (PER), 1998 p.9). Unemployment rates in Rosario during the 90s reached 20% (Plan Estratégico Rosario (PER), 1998) being worse among vulnerable groups and young people. The socio-economic and political crisis that outburst in Argentina in December 2001 increased the percentage of people living under the poverty line in Gran Rosario .
Table 1: Percentage of people living in households under poverty line. May 2002. Source INDEC- Household Survey: “Encuesta Permanente de Hogares”
The proportion of households with an income inferior to acquire a basket of food and services rapidly grew to 61 % in May 2003 (See: Table 1). Within the four most populated metropolitan areas of Argentina, Great Rosario is the one with the largest percentage of its population living under the poverty line. This can be explained by the (historical) presence of an important number of slums. In the second half of the twentieth century, due to urban and rural migration (caused mainly by the deterioration of regional economies), the number of informal settlements (“villas de emergencia”) started to grow. In 1996, The Municipal Bank Foundation (Fundación Banco Municipal, 1996) surveyed 91 slum settlements inhabited by a total of 22,685 families (113,382 inhabitants, 11% of the total population of Rosario). It is possible to appreciate the vulnerability to social exclusion of its inhabitants, since 50% of the slums population over 15 years old has not completed the primary school (Plan Estratégico Rosario (PER), 1998). The unequal characteristics of its population are reflected in the different levels of housing qualities and access to services (infrastructure and social services).
At the same time, the number of gated communities has increased during the 90s, causing an increase in housing inequalities, disparities and social segregation. Bragos, (2001) explains that even though the first gated communities were built close to low-income residential areas, in the last years developers are trying to “clean” the land for new investments and (low-income) families are moving out looking for cheaper land sold by the same developer. According to the same author this generates the formation of clusters of homogeneous groups (social condition, education, age, family type, and ideal way of life) (Bragos, Mateos et al., 2001).
The previous paragraphs illustrate the unequal levels of quality of life for a high percentage of the population of Rosario. Besides, the degree of vulnerability that they are exposed to, it illustrates the importance of implementing a methodology to monitor and evaluate inequalities to support effective intervention policies.
Considering that the use of GIS-based indicators should be policy oriented, the unit of analysis of this research is delimited within the administrative boundaries of the city. Rosario falls within the context of a city both under the influence of globalisation (privatisation of services, deregulation, etc.) and a local administration that implemented a decentralization process expressing the willingness to improve the welfare level of the people living in its districts and to reduce disparities.
Planning context, a decentralization process which demands more area-based information
During the second half of the 1990s, the city of Rosario started a process of modernization of its local administration and the implementation of new planning tools such as the Strategic Plan of Rosario (PER) and the “Decentralization Programme”.
In 1995, the Municipality of Rosario started the decentralization processes which led to the creation of six districts. The intention to reduce inequalities has been explicitly stated in the objectives of the Decentralization Programme and in article 5.1 of the plan of actions, the importance of comparing different geographical areas is justified “to adjust equity criteria in the assignment of resources”. This decentralisation processes goes hand in hand with new urban management models and the re-emergence of a stronger role for local governments which appeared in the late 1980s supported by many donor agencies (Devas and Rakodi, 1993). These decentralization processes that are taking place in different parts of the developing and developed world generate an increase of demand of information at lower levels of detail such as neighbourhood level.
Since 1980 the welfare institutions which depended on the central government have been dismantled (Bifarello, 2000). A process of decentralization makes provincial and municipal governments responsible for the provision of those services. Decentralization has brought decision-making closer to the people, and this is very much reflected in the local decentralization process that took place in Rosario. On the other hand, the withdraw of the central government in the provision of social services but with a lack of enough financial resources put a lot of pressure on local authorities which are not always willing or able to accept the new responsibilities (Bifarello, 2000). This requires local governments to focus on differences within their territories and represents a great opportunity to introduce GIS into policy making by targeting those differences.
Case study: Describing intra-urban inequalities with GIS-based indicators.
According to Innes (1990) some of the problems faced by the early social indicators movement in the 1960s were that they emphasized the measurement task, often excluding political and institutional aspects. To avoid the same problems, GIS-based indicators should be easily understood and transparent to planners and decision makers as well as related to the local policy context.
Twenty semi-structured interviews were held in the city of Rosario with the objective to find out how local policy makers perceive inequality as a problem to solve; and to identify which domains/factors of inequality they find more relevant. These two objectives are very much related to a valid selection of indicators.
The interviews included the ten members of the municipal cabinet and the six directors of the decentralized districts covering key decision makers in the local government. To gain more specialized perspectives it was also decided to extend the interviews to the directors of the Public Housing Service, the Strategic Planning Office (PER), the Origin and Destination Survey Project (Public Service Secretary), and the Decentralization Program.
The interviews were of the semi-structured type, that is, the interviewee was asked closed questions from a pre-printed list as well as open questions; and it was followed by a more informal dialogue (open-ended type interview). The advantage of this approach is that comparisons can be made between the different answers as well as incorporating extra observations and opinions not contemplated in the pre-defined questions.
The pre-selection of indicators included in the interviews was based on literature review of indicators (UNCHS, 1995; OECD, 1997; UNCHS, 2000; UNCHS, 2000) and they were adjusted according to the spatial inequality domains that this research intends to address.
The interviewees were given a questionnaire with a list of 13 aspects related to inequality (See: Table 2). They were asked to indicate how important they consider each aspect applying a Likert scale (1=Very important; 2=Important; 3=Neutral; 4=Unimportant; 5=Very Unimportant). The aspects were grouped in three main classes: aspects related to the household and its members, aspects related to the dwelling and aspects related to accessibility.
Table 2: Aspects of inequality
From the interviews, it was found that the five most important aspects of inequality, according to the policy makers were: overcrowding, education, employment, water connections and accessibility to schools. This selection might be explained by the notion that most of the policy makers shared similar values about basic needs or basic rights. The choice of the aspects education and employment might be explained by the agreed idea among policy makers that through better education, skills and employment, economic and other disparities can be levelled. Both of them are also social aspects related to the households; while overcrowding and water connections (tap water inside the house) are two dimensions related to the dwelling. Where overcrowding shows the mismatch between the housing needs of the different households types and the house size (or number of bedrooms), water connections shows the capacity of the household to access one of the most important basic needs.
To know how relevant it was to measure a specific inequality aspect and how it can be improved locally, the interviewees were asked whether the local government is able to reduce the level of inequality. The affirmative responses for each aspect are shown in Table 3. There is a strong agreement between the interviewees that the inequality aspects related to the household and its members (e.g. income) cannot be directly influenced locally. There is a variation between the aspects related to the dwelling and the infrastructure provision which can be explained by the fact that some of the services are privatised (e.g. water and sewage). Finally, there is a strong agreement that the accessibility aspects (e.g. to schools) can be influenced locally. The average importance given to the different aspects was compared to the percentage of policy makers that agree that the local government is able to intervene at local level. From Table 3 it can be analysed that from the first five ranked aspects only overcrowding and accessibility to schools have more than 50% of the interviewees agreeing that they can influence them locally.
Table 3: Level of importance of inequality aspects (5 point scale, 1 = highest priority)) and percentage of policy makers that agreed it is possible to influence them locally
Most of the inequality aspects were seen as highly interrelated, especially unemployment, education, health and water and sewage infrastructure. Although there are limitations in the possibility to influence some of the inequality sources such as income and employment, it seems to be possible to tackle its consequences and improve the quality of life for those affected by inequality. In spite of the implementation of social programmes, it was expressed that they are not able to reduce certain aspects of inequality directly either because is the responsibility of a higher administrative level (e.g. income or education) or a privatised public service (e.g. water or sewage). However, to reduce inequality in those aspects they can influence or lobby at a higher government level, which justifies the presentation of objective reports elaborated with indicators (some examples were mentioned in that respect).
Present use of indicators.
The second part of the interviews concentrated on the objective to identify the usage (or not) of indicators by policy-makers on decision-related issues and to explain the reasons.
To determine in which part of the policy cycle they have been using indicators and when they would like to use them in the future, a questionnaire was provided to the interviewees. It was asked to rank the following phases according to the degree of use of indicators. They ranked them twice, first considering how they had been using up to now and secondly how they would like to use them in the future:
Diagnose and problem identification was seen as the most important phase where indicators were used as well as where they would like to use them in the future.
During the interviews, it was possible to explain some of the reasons why in some cases they have not used indicators. The main topics mentioned by the interviewees were the lack of staff and/or experience, lack of funds and time to quantify. Remarkably, most of them agree that it is very important to quantify and use indicators in the different policy cycle phases but they refer to the tendency to replace quantification studies and planning by intuition or local knowledge. Some justify this by the need to act urgently in a very unstable context. On the other hand, it was possible to find some examples of present use of indicators.
Supranational organisations have a role in the demand of, or in the collection of indicators as well as in the set of targets. In Argentina the IMF, the World Bank or the Inter-American Development Bank make local governments use indicators to monitor progress/performance and target areas. Another reason why indicators can be implemented is to control the performance and the implementation of privatised services.
Concern was expressed about the lack of planning tradition with as a consequence less interest in measuring. In any case, there is high recognition by policy makers of the importance of having the best information possible when it comes to take decisions or start implementing a programme or action. Besides, most of the interviewees recognised that they rely not only on quantitative but also on qualitative information and they highly appreciate the combination of both.
In terms of data availability, policy makers recognise the need to work with small areas but they are aware of the difficulties in finding data at such a high level of disaggregation. The economic difficulties that affected the schedule in the implementation of the Census 2001 in Argentina are a clear example of that. According to one interviewee the lack of information does not seem to be a problem but the lack of systematically presented information. Other interviewees considered that quantifying is important but still there is a lack of personnel working on that. Other reactions expressing the same concern:
“The teams we have to perform the tasks are very limited and sometimes they don’t have the technological resources. In our case the staff share one PC and this is very limiting because they take turns to use it.”
“Most of the areas can be approached at a micro scale but for that we require staff”
One interviewee brought the issue that it is not a problem of lack of staff, resources or know how but a problem of “lack of planning culture”:
“There is always a point in the development of the administration that it is necessary to evaluate, to correct and improve. That [evaluation], in general, it is not being incorporated and things are mostly done in an arbitrary and non-rigorous way. That is: it functions essentially based on demand. This is also related to the lack of planning culture”.
It was stated in the interviews that even though they have used indicators they are “still in the search of how to improve the link between indicators and policy decision making” The lack of use of indicators in decision-making was mentioned by other interviewees:
“There is not a practice of using these elements [indicators] when it comes to decision making and in general these decisions are more intuitive or empiric rather than based on information.”
One positive aspect was that the multidimensionality of inequality (and social problems in general) makes policy makers act in cooperation with different areas and secretaries and seem to make them use more indicators. Some of the interviewees also mentioned that due to the need to receive credit from an international agency they started to cooperate with another area within the local government and indicators where presented and exchanged to prepare the reports: “To determine the neighbourhoods [that will receive a credit for urban art] the Inter-American Development Back asked a socio-economic study of these neighbourhoods”.
GIS- based indicators constructed from census data.
After the interviews, it was decided to include the ten most important indicators suggested by the policy makers as a representative set of indicators of aspects of inequality. Income level is not included in the set since there is no income data available but a socio-economic status index is used as a proxy and constructed with factor analysis. Accessibility to internet, although it was considered the lowest important aspect of inequality, is still kept since it is necessary to cover the domain of the accessibility to ICT and new technologies considering its future implications in the quality of life of individuals and that technological diffusion is selective (Castells, 1996). The accessibility to day-care centres (“Centros Crecer”) was included since they represent a policy response to inequalities. Expressed housing needs, calculated from administrative data, are also included. This indicator is also relevant due to the high importance of the habitat in the quality of life of the population and is included as an example of how administrative data can also be used to construct GIS-based indicators.
Table 4: Final set of indicators
In the explorative study, reported on in this paper, indicators are constructed for the five most important aspects according to policy makers:
These five indicators can be considered as the top priorities according to the policy makers and they cover all the inequality domains.
With the use of a desktop GIS (ArcGIS) a selection of 2001 census variables was used to construct and map the five selected indicators. The spatial detail of the census data is determined by the boundaries of each census tract (“fracción censal”). They are established so that the population in each is approximately of 13,500 people, regardless of its geographic size. The National Institute of Statistics and Censuses from Argentina (INDEC) divides each census tract into 15 smaller units called “radio censal” with a population of approximately 900 people. These “radios censales” are built up from individual blocks (“manzanas”) but the data is not available to the public at that level of disaggregation.
Figure 1: Block groups ("radios censales")
As shown in Figure 1, tracts are formed of an average of 15 block groups (“radios censales”). There are 901 Blocks groups in Rosario with and average of 1015 People in each. The following are the resulting GIS-based indicators constructed at block group level. The best off areas are shown in white and the worst off areas in black. A spatial patterning of inequality can be found in the city of Rosario with a concentration of needs in certain areas (See Figures 2 and 3).
Figure 2: Spatial patterning of inequalities in Rosario. Source: own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina)
Figure 3: Accessibility to kindergartens and primary schools (minimum distance from census tract gravity point, in meters) Source: Own analysis
If we analyse the indicators of the best off block group and the worst off block group and between the best 10% and worst 10% it is possible to appreciate the considerable gap between both groups (Table 5).
Table 5: Gaps between best off and worst off areas
The gaps between these two extreme groups are noticeable; for example, the percentage of overcrowding affecting households in the worst off 10% of the block groups is 20 to 50 times greater than in the best off decile. Unemployment can affect up to 50% of the population of the worst off areas compared to only 4% to 8% in the best off area.
Aggregation level – Ranking neighbourhood areas
Considering that indicators should be related to policy actions, it is important to see how inequality is depicted at a level of aggregation coincident with the area of intervention of policy actions. In many cities this is the case for the municipality districts. Policy makers might preferably act at district level where neighbourhood participation activities also take place at that level. However, the definition of neighbourhood areas to analyse the data is quite controversial since we come across the problem of agreeing on a common definition of neighbourhood. Not even citizens agree on the boundaries of the neighbourhood. There is no clear boundary acceptance since the different perceived neighbourhoods overlap in space and sometimes the inhabitants even have alternative names and borders.
In the case of Rosario every district is subdivided by the Municipality into six “areas barriales” or neighbourhoods areas. To provide policy makers with an option of ranking those areas, the indicators constructed at block group level were summarised within ArcGis into neighbourhood areas and ranked from the worst off to the best off (See Figure 4 and Table 6)
Figure 4: Neighbourhood areas with at least one census block ranked within the five worst off areas in terms of overcrowding, unemployment, illiteracy rate and tap water inside the house.
Table 6 : Neighbourhood areas with at least one census block ranked within the five worst off areas in terms of overcrowding.
The influence of the areal unit definition in the description of inequalities is very relevant due to the scale factor. Previous research on segregation that carried out comparative studies between European cities (Kruythoff, 1998) can show how the areal analysis unit (e.g. municipality, district, neighbourhood) and the average population per unit influence the appreciation of the phenomenon. As it can be expected, spatial differences might be smaller in larger areal units of analysis. Initiatives to gather indicators at neighbourhood level are taking place and they agree that differences within the city are so marked that averaging social indicators at district level leads to wrong diagnoses and policy allocation mistakes (Kingsley Thomas G., 1999). Hence using indicators at higher levels of aggregation can be misleading if this problem is not contemplated. As it can be seen in Table 6 when the indicator Overcrowding is aggregated into neighbourhood area level the difference between the minimum and maximum values can be quite significant, which is also reflected in the standard deviation.
GIS-based indicators constructed from administrative data – “Hidden need”
One of the problems of collecting indicators at intra-urban level is the lack of data other than census data –most of the research done on spatial inequality is based on this source-. This poses a conflict to solve, which is to find alternative sources (other than census data). With this in mind, this research starts addressing this issue by including the use of points of expressed demand (Martinez, 2000). Indicators from census data are good to measure indirect need (or demand) but they cannot measure expressed demand coming from the population. While studying the role of GIS in urban planning, Webster (1993) identified expressed demand as a component of descriptive analysis. Webster differentiates between imputed and expressed demand. Imputed demand (or derived demand) is evaluated indirectly by inference from locational information by means of demand indicators; on the other hand, expressed demand is evaluated directly by recording the expressions of demand by members of the public.
In the case of housing needs, for example, needs can be evaluated indirectly by means of indicators. On the other hand, expressed demand/need is evaluated directly by recording the expressions of demand by citizens. The number of registered demands for good housing is also considered a good indicator of the unsatisfied needs and may serve as an alternative indicator (Mega, 1995).
Administrative databases available at the municipalities are very useful to describe inequalities at a low level of detail where census data is not available and as a complimentary data source. In this research data provided by the Public Housing Office (SPV) is used. This database contains data of 12,926 cases of express demand for housing solutions. They can also be aggregated at block or census tract level if needed.
To detect where the expressed demand is coming from, a point map is produced. The data used to elaborate this map was tabular data created at SPV by the Social Work Department. The field selected was that of the address of the person that voluntarily went to the SPV office and asked for a housing solution or expressed his housing problem.
Geocoding and address matching were the GIS processes used to map the expressed demand, where the address provides the locational key of the demand. The address matching capability of ArcGIS can generate point locations that represent events on a map containing (like in this case) a street network ready for geocoding. The map network has address ranges for the right and left sides of the streets assigned to the segments. ArcGIS reads the address from the event tabular data and estimates a map position. After performing the geocoding process, 76% of the addresses were matched.
This approach proved to be efficient to detect cases of housing needs where derived demand via indicators is showing lower levels of demand or hiding the situation at all. This can help in the detection of the “new poor” normally hidden by many indicators . This new poverty is scattered throughout the cities and since they are not living in recognisable poor neighbourhoods they can be found in any middle-class apartment block (Minujin, 1995)
In Figure 5 we can see how points of individuals expressing their housing needs were “hidden” in an area classified as within the 10% best off block groups. At the same time, if we analyse the neighbourhood area (on the left map) we can see how two extreme realities are close to each other: there is a block group (hatched polygon) which belongs to the best 10% areas only 400 m to a block group which belongs to the worst 10% area (grey polygon).
Figure 5 : Neighbourhood areas with contrasting realities. Own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina) and SPV. Expressed demand of housing (each dot represents the demand from one family).
Factor analysis
Finally, a more exploratory approach was chosen to compare the five most important aspects of inequality indicators (selected after the interviews and the literature review) with the factors obtained with factor analysis. Factor analysis derives from a large set of variables, a set of factors that can be thought of as “super variables which represents a cluster of highly correlated census-based variables” (Davies 1984 in: Pacione, 2001). The meaning of each factor can be associated by the original variables with high loadings.
To generate the factors an extended set of variables obtained from the 2001 census was analysed. A further selection of variables was done by discarding those that had low communality and those that generate a “closed system” (e.g. households with cell phone and households without cell phone). A list of 57 variables was finally included to extract the factors using principal components and a rotated (orthogonal) solution (which means that the factors do not correlate with each other but optimally with the common variance in the original variables). From the factor analysis four factors were identified and named according to the variables that contain.
Factor 1: Wealth and Social Status; (shown in the following map)
Factor 2: Family (age);
Factor 3: Family (status - woman headed households);
Factor 4: Housing (inadequacy or housing deprivation).
From this result, we can suggest to consider Factor 1 as a good proxy of income inequality since it is highly related to wealth and social status, with high loadings on variables such as ownership of cell phone, PC with internet connection, cable TV, employment, university level education and private school attendance.
As we concluded in the previous section, if we compare the wealth factor with the points of express demand we can find “hidden” needs in areas classified as wealthy (Figure 6). It is clear that to avoid the problem of ecological fallacy we should not assume that every person living in a wealthy area is wealthy.
Knox and Pinch (2000 p.123) also warn about the dangers of ecological fallacy (i.e. making inferences about individuals with data based on aggregates of people) and put it in this way
“not everyone in a deprived area is necessarily deprived and not every deprived person in an area of ‘multiple deprivation’ is necessarily multiply deprived”.
Figure 6: Own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina) and SPV. Expressed demand of housing (each dot represents the demand from one family).
Conclusions
In 1995, the Municipality of Rosario started a decentralization processes which led to the creation of six districts. They stated the importance of comparing different geographical areas “to adjust equity criteria in the assignment of resources”. As many other local governments around the world, they are faced to the importance of having good information to support policymaking. In this paper, the use of GIS-based indicators was applied in a case study where they prove to be valuable to detect need areas and communicate detailed geographical patterns of inequality. Here it is suggested that GIS has a good opportunity to be integrated into policy making via indicators that are linked to real problems and policy goals (in this case: “to adjust equity criteria in the assignment of resources” or to target need areas).
Some advantages of GIS to construct indicators emerge from this case study. As it was explained in the first section to operationalize indicators, it is necessary to: organize data, to quantify and to communicate. In this case, it was possible to integrate different data sources such as census and administrative data, quantify needs and analyze the gaps between best and worst off areas, and to generate maps to communicate and detect problem areas.
Policy makers confirmed the need to use indicators in a more systematic way and mostly to use them during the problem description phase of policy making. However, many of those interviewed expressed their concerns about issues such as lack of data (or systematized data), lack of staff and/or resources. The lack of a tradition of the use of quantitative analysis within the policy context of Rosario might be improved by the provision of specific information attached to policy goals and specific phenomena with GIS-based indicators. The multidimensional aspect of inequalities and the demand to target need areas with indicators is a good opportunity both to induce the exchange of information between different municipality departments and to implement the use of GIS in policy making.
Beyond this, it is necessary to improve the provision of small area information and a better coordination between municipality departments to generate consistent administrative data that are suitable both for the construction of indicators and for the generation of geographic data.
References
The Integration of GIS into Policy Making through Intra-Urban Indicators, Case Study Rosario (Argentina)
Javier A. Martínez MSc.
Urban and Regional Research Centre Utrecht (URU)
Faculty of Geographical Sciences - Utrecht University. Utrecht.
International Institute for Geoinformation Science and Earth Observation (ITC)
Netherlands.
Javier A. Martínez MSc.
Urban and Regional Research Centre Utrecht (URU)
Faculty of Geographical Sciences - Utrecht University. Utrecht.
International Institute for Geoinformation Science and Earth Observation (ITC)
Netherlands.
Abstract
Policy makers are currently encouraged to introduce area-based policies to target deprived areas, then set priorities, and reallocate resources. GIS-based indicators can be a valuable tool to describe differences in the quality of life and access to services and identify trends. GIS-based indicators provide valuable information especially if they are policy related, and to some extent they are able to generate better decisions and evaluate the policy performance. This paper explains how the integration of GIS into policymaking can be achieved using intra-urban indicators and shows through a case study the advantages of GIS in constructing them. GIS-based indicators are constructed combining different data sources such as census and administrative data. This paper also analyses how policy makers perceive the problem of urban inequalities and how they deal with data and indicators on decision-related issues.
To succeed in the GIS adoption and capacity building in developing countries, GIS should be able to respond to the local needs and it should be policy demand driven and application context sensitive. The use of intra-urban indicators can contribute to the integration of GIS into policymaking as well as meet local problems and needs.
Introduction
By means of a case study this paper explains how the integration of GIS into policymaking can be achieved using intra-urban indicators. This section introduces the problem of inequality within cities and the growing demand of information to target disadvantaged areas. The second section describes what indicators are, their functions in policy making and why GIS can help in its operationalization. The third section deals with the case study area and its planning context where the methodology to describe the inequality phenomena with GIS-based indicators is applied. The fourth section starts by analysing the inequality aspects that policy makers consider more relevant and how they deal with indicators. Finally, it is shown the operationalization of a set of indicators and how they can help to identify gaps and rank areas.
The “Global Urban Observatory”, the UN-HABITAT international capacity building network, indicates that many cities suffer from an information crisis that undermines their capacity to develop effective urban policy. It also warns that these cities do not have a sustained or systematic approach to assess the urban problems and cannot evaluate the success of the implemented policies. Urban indicators are seen as a tool that can improve that situation (Moor, 2000). There is also a recognition that GIS can be used for the collection and analysis of urban indicators. A proof of that appreciation is that in February 2003 ESRI donated 15 million dollars to the Global Urban Observatory’s indicators programme .
Since the Agenda 21 declaration, the importance for a sustainable development of reducing inequalities and disparities within cities it is constantly mentioned (United Nations, 1992; European Commission, DG XI (1994) in Mega, 1996; UNCHS, 2001). Spatial inequality occurs in urban areas around the world. However, especially in cities in developing countries, inequalities in the habitat conditions or access to social and physical infrastructure are particularly evident. At the same time, local governments are encouraged to target those deprived areas. Problems related with deprivation or poverty show spatial concentrations in cities accentuating the problems suffered by people living in certain areas (Pacione, 2001 p. 291).
Socio-territorial indicators are being used since the beginning of the twentieth century. In the 1970s, there was a growth in studies of patterns of inequality and spatial injustice with an interest to influence public policy. Later on, postmodernism emphasized or “celebrated” diversity and difference (Smith, 1994). In the 1990s there was again a growing engagement of geographers with inequalities, moral and social issues, including the theme of the ethics of professional practice (e.g. Couclelis, 1999). This growing concern is particularly evident and it is reflected in different reports and initiatives of international organizations that are stressing the importance of monitoring spatial inequalities within cities (e.g. UNCHS, 1995; The World Bank, 1996; EC (European Communities), 2000; European Commission, 2000; UNDP, 2000; UNCHS, 2001; UNDP, 2001).
Even though there is growing literature about inequalities and attention for the importance of measuring it; there is not enough discussion about methods and tools (basically indicators), how suitable they are, and the use of GIS to operationalize them. This concern is also expressed as a problem of lacking of “spatially relevant indicators” (Kunzmann, 1998). Several studies of patterns of inequality and spatial differentiation within the city have been based on indicators of personal income (e.g. Smith, 1973; Smith, 1994; Chakravorty, 1996). However, as Knox and Pinch (2000 p.100) put it “While socioeconomic differentiation is arguably the most important cleavage within contemporary cities, it is by no means the only one”. Besides, for most cities there is also a data availability restriction, since data of personal income is not collected at low levels of aggregation. This is the case of many cities in the developing world as well as in developed countries (see: Broadway and Jesty, 1998; Wessel, 2000).
Some of the criticism to “area-based” policy strategies to reduce inequalities argues that targeting is not based on needs and that the areas are not homogeneous (the ecological fallacy problem). To minimize the problems (and criticism) of ecological misleading outcomes, this research focuses on the use of small area units and low levels of aggregation and places stress on “need” more than any other distributive criteria. Furthermore, the construction of indicators outside a policy perspective is also widely criticised (Innes, 1990). Consequently, this research proposes to link the development and use of GIS-based indicators with policy.
The city of Rosario (Argentina) has different levels of housing needs and access to physical and social infrastructure. Rosario is taken as a case study and GIS-based indicators were used to measure spatial inequalities. Census data from 2001 was analysed together with
administrative data from the Public Housing Office. GIS-based indicators provide valuable information particularly if they are policy related and able to generate better decisions. For that reason, through a series of interviews it was analysed how policy makers perceive the problem of urban inequalities and how they deal with data and indicators on decision-related issues.
Overcrowding, education level, employment, tap water inside the house and access to primary schools were the five most important aspects of inequality according to policy makers. Diagnose and problems identification was the phase within the policy cycle where they would more appreciate the use of indicators. After the analysis, it was possible to identify the most disadvantaged neighbourhood areas and to measure the gap between the best off and worst off areas. Needs derived from indicators constructed from census data were complemented with point maps showing directly expressed need which proves to show “hidden” needs not captured by the derived indicators. Finally, factor analysis was applied to define a socio-economic factor which can be used to describe the socio-economic gaps in the city and as a proxy indicator of income.
Policy makers are currently encouraged to introduce area-based policies to target deprived areas, then set priorities, and reorient resources. GIS-based indicators prove to be a valuable tool to describe differences in the quality of life and access to services. To succeed in the GIS adoption and capacity building in developing countries, GIS should be able to respond to the local needs and be policy demand driven. The use of intra-urban indicators can contribute to the integration of GIS into policymaking as well as meet local problems and needs.
What are indicators and why GIS can help in its operationalization?
Indicators simplify complex phenomena into quantifiable measures that can be used for policy and decision-making. Therefore, indicators have three functions: to simplify, to quantify, and to communicate (Van Delft, 1997). Innes (1990) considers that an indicator focuses and renders intentionally selected areas of the reality. She puts it in this way: “An indicator is simply a set of rules for gathering and organizing data so they can be assigned meaning”
From these definitions, we can see the potential that GIS has to operationalize indicators. While constructing indicators, it is necessary to: organize data, to quantify and to communicate. These three indicators functions coincide with the usually acknowledged advantages of GIS: data organisation, spatial analysis and visualisation (Burrough, 1986 in: Huxhold, 1991; Webster, 1993; Ghose and Huxhold, 2002). Furthermore the spatial dimension of urban inequalities and the area-based policies that target deprived areas makes decision support suitable for analysis and monitoring with the use of GIS based indicators. Besides, in the late 1990s the value and potential of GIS to construct intra-urban indicators is increased by a combination of: a growing concern on intra-urban inequalities, the implementation of area-based policies and the developments in ICT and GIS technology.
Apart from rendering intentionally selected areas of the reality (Innes, 1990) GIS-based indicators can perform different functions which are related to the different phases in the policy cycle. Indicators can be classified as follows :
- Descriptive or Baseline indicators: used to present a certain state or process. They describe the present situation and are useful to target and identify need areas. They usually represent the starting point before policies are applied.
- Normative or Target indicators: used to evaluate and compare the result of an indicator to a standard that acts as a norm (usually related to policies). With these indicators goals and objectives can be set after the areas of need have been identified with the descriptive/baseline indicators.
- Performance or Outcome indicators: used to present achievements by measuring whether the policy goals have been achieved as well as the level of people’s satisfaction with the results.
GIS-based indicators to measure spatial inequality
GIS can be used to operationalize indicators in any of the policy cycles. Besides, it should be established which policy goal is considered and which phenomenon the indicators will describe. In this case study, indicators are used to describe spatial inequality and can be related to the policy objective of targeting intra-urban inequalities.
In this research and to justify that inequalities in cities really matter it is necessary to consider an approach from a social justice perspective. When there is a need to monitor or describe inequalities, it is implied that with the use of planning tools there is the intention to change, improve and/or solve this problem. Hence, there is a concern with what the city should be, and some form of redistributive or compensatory action. According to Smith (1994 p.2) “questions of social justice, morality or ethics are usually described as normative, concerned with what should be, as opposed to positive knowledge which is about what actually is”. Smith (1994) considers that justice involves treating people fairly, which in distributive justice means that whatever is being distributed should go to people in the right quantities. He expresses that “fairness” means that people in the same circumstances should be treated in the same way. A difference can be made between arithmetic equality and proportional equality. In arithmetic equality everyone is getting exactly the same quantity of something and in proportional equality the distribution is justified according to a certain criteria such as need or market demand. In this research a need approach within a social justice perspective is taken.
The spatial patterning of inequality and difference results in the segregation of certain segments of the population (Langlois and Kitchen, 2001). Actions against that problem are usually addressed with area based policies that target those deprived or segregated areas. Social justice it is concerned with the question of who gets what where and how, and more precisely who should get what where and how (Smith 1977 in: Pacione, 2001). We can clearly see from these definitions that the spatial aspect of inequalities justifies the use of GIS-based indicators.
In this research, spatial inequality is considered to be a heterogeneous, multidimensional and complex phenomenon with several aspects. Two axes are distinguished with the following domains:
(un)equal conditions of quality of life
- social environment
- physical environment
- social infrastructure
- physical infrastructure
- virtual infrastructures (nets, ICT- “switched off areas”-)
(un)equal distributions of opportunities / (un)equal access to
Criteria for constructing GIS-based indicators
After establishing the domains that cover the inequality phenomena, it is necessary to select a particular set of GIS-based indicators. The selection is based both on the selected domains and on effectiveness and efficiency criteria. The following UN-HABITAT criteria (UNCHS, 1995 p.18) were considered:
- Importance for policy Indicators should be directly relevant to existing or proposed urban or shelter policy, and should directly measure outcomes.
- Easily understood Simple indicators that can be understood by those without specialist knowledge are likely to have a far wider currency and interest and to be used more accurately and readily.
- Cost-effective and timely Indicators should be able to be collected in a cost-effective way and on a regular basis which reflects the rate at which the indicator is expected to change.
- Include most disadvantaged Where equity is a concern, indicators should focus on the most disadvantaged rather than the whole income distribution.
- Available for geographical areas or social groups Indicators which can be disaggregated are likely to be of greater interest and are likely to be used in a wider variety of circumstances; indicators should always be disaggregated by sex, age and geographic area where special needs and equity are policy issues.
Following these criteria and both to relate the selection of indicators to local policy, and to integrate GIS into policymaking, in this research the selection of indicators was derived from present policy interests in relation to inequality. For that reason, interviews with policy makers were carried out. To be able to construct geographically disaggregated indicators two data sources were selected: census data and administrative data. The study area for the application of the methodology is the city of Rosario (Argentina).
Case study area characteristics and planning policy context, Rosario (Argentina)
Rosario is the third largest city in Argentina and it is located along the shores of the Paraná River at approximately 300 km northwest of Buenos Aires.
Rosario is the main city within a metropolitan area which has a population of around 1,300,000 inhabitants (“Gran Rosario”). The city of Rosario itself, according to the 2001 census, has a population of 923,424 persons living in an area of nearly 180 km2. In 1970s a crisis affected the economic structure of the region (Plan Estratégico Rosario (PER), 1998). As in many other cities around the world, Rosario suffered the effects of deindustrialization. The adjustment and reconversion of some industries (e.g. iron, paper, steel and chemical industries) “put the region in a critical economic and social situation, with difficulties to compete in a frame of open economy context and international competition” (Plan Estratégico Rosario (PER), 1998 p.9). Unemployment rates in Rosario during the 90s reached 20% (Plan Estratégico Rosario (PER), 1998) being worse among vulnerable groups and young people. The socio-economic and political crisis that outburst in Argentina in December 2001 increased the percentage of people living under the poverty line in Gran Rosario .
At the same time, the number of gated communities has increased during the 90s, causing an increase in housing inequalities, disparities and social segregation. Bragos, (2001) explains that even though the first gated communities were built close to low-income residential areas, in the last years developers are trying to “clean” the land for new investments and (low-income) families are moving out looking for cheaper land sold by the same developer. According to the same author this generates the formation of clusters of homogeneous groups (social condition, education, age, family type, and ideal way of life) (Bragos, Mateos et al., 2001).
The previous paragraphs illustrate the unequal levels of quality of life for a high percentage of the population of Rosario. Besides, the degree of vulnerability that they are exposed to, it illustrates the importance of implementing a methodology to monitor and evaluate inequalities to support effective intervention policies.
Considering that the use of GIS-based indicators should be policy oriented, the unit of analysis of this research is delimited within the administrative boundaries of the city. Rosario falls within the context of a city both under the influence of globalisation (privatisation of services, deregulation, etc.) and a local administration that implemented a decentralization process expressing the willingness to improve the welfare level of the people living in its districts and to reduce disparities.
Planning context, a decentralization process which demands more area-based information
During the second half of the 1990s, the city of Rosario started a process of modernization of its local administration and the implementation of new planning tools such as the Strategic Plan of Rosario (PER) and the “Decentralization Programme”.
In 1995, the Municipality of Rosario started the decentralization processes which led to the creation of six districts. The intention to reduce inequalities has been explicitly stated in the objectives of the Decentralization Programme and in article 5.1 of the plan of actions, the importance of comparing different geographical areas is justified “to adjust equity criteria in the assignment of resources”. This decentralisation processes goes hand in hand with new urban management models and the re-emergence of a stronger role for local governments which appeared in the late 1980s supported by many donor agencies (Devas and Rakodi, 1993). These decentralization processes that are taking place in different parts of the developing and developed world generate an increase of demand of information at lower levels of detail such as neighbourhood level.
Since 1980 the welfare institutions which depended on the central government have been dismantled (Bifarello, 2000). A process of decentralization makes provincial and municipal governments responsible for the provision of those services. Decentralization has brought decision-making closer to the people, and this is very much reflected in the local decentralization process that took place in Rosario. On the other hand, the withdraw of the central government in the provision of social services but with a lack of enough financial resources put a lot of pressure on local authorities which are not always willing or able to accept the new responsibilities (Bifarello, 2000). This requires local governments to focus on differences within their territories and represents a great opportunity to introduce GIS into policy making by targeting those differences.
Case study: Describing intra-urban inequalities with GIS-based indicators.
According to Innes (1990) some of the problems faced by the early social indicators movement in the 1960s were that they emphasized the measurement task, often excluding political and institutional aspects. To avoid the same problems, GIS-based indicators should be easily understood and transparent to planners and decision makers as well as related to the local policy context.
Twenty semi-structured interviews were held in the city of Rosario with the objective to find out how local policy makers perceive inequality as a problem to solve; and to identify which domains/factors of inequality they find more relevant. These two objectives are very much related to a valid selection of indicators.
The interviews included the ten members of the municipal cabinet and the six directors of the decentralized districts covering key decision makers in the local government. To gain more specialized perspectives it was also decided to extend the interviews to the directors of the Public Housing Service, the Strategic Planning Office (PER), the Origin and Destination Survey Project (Public Service Secretary), and the Decentralization Program.
The interviews were of the semi-structured type, that is, the interviewee was asked closed questions from a pre-printed list as well as open questions; and it was followed by a more informal dialogue (open-ended type interview). The advantage of this approach is that comparisons can be made between the different answers as well as incorporating extra observations and opinions not contemplated in the pre-defined questions.
The pre-selection of indicators included in the interviews was based on literature review of indicators (UNCHS, 1995; OECD, 1997; UNCHS, 2000; UNCHS, 2000) and they were adjusted according to the spatial inequality domains that this research intends to address.
The interviewees were given a questionnaire with a list of 13 aspects related to inequality (See: Table 2). They were asked to indicate how important they consider each aspect applying a Likert scale (1=Very important; 2=Important; 3=Neutral; 4=Unimportant; 5=Very Unimportant). The aspects were grouped in three main classes: aspects related to the household and its members, aspects related to the dwelling and aspects related to accessibility.
To know how relevant it was to measure a specific inequality aspect and how it can be improved locally, the interviewees were asked whether the local government is able to reduce the level of inequality. The affirmative responses for each aspect are shown in Table 3. There is a strong agreement between the interviewees that the inequality aspects related to the household and its members (e.g. income) cannot be directly influenced locally. There is a variation between the aspects related to the dwelling and the infrastructure provision which can be explained by the fact that some of the services are privatised (e.g. water and sewage). Finally, there is a strong agreement that the accessibility aspects (e.g. to schools) can be influenced locally. The average importance given to the different aspects was compared to the percentage of policy makers that agree that the local government is able to intervene at local level. From Table 3 it can be analysed that from the first five ranked aspects only overcrowding and accessibility to schools have more than 50% of the interviewees agreeing that they can influence them locally.
Present use of indicators.
The second part of the interviews concentrated on the objective to identify the usage (or not) of indicators by policy-makers on decision-related issues and to explain the reasons.
To determine in which part of the policy cycle they have been using indicators and when they would like to use them in the future, a questionnaire was provided to the interviewees. It was asked to rank the following phases according to the degree of use of indicators. They ranked them twice, first considering how they had been using up to now and secondly how they would like to use them in the future:
- Diagnose and problems identification;
- Policy and program formulation;
- Implementation and goals establishment;
- Evaluation and correction.
Diagnose and problem identification was seen as the most important phase where indicators were used as well as where they would like to use them in the future.
During the interviews, it was possible to explain some of the reasons why in some cases they have not used indicators. The main topics mentioned by the interviewees were the lack of staff and/or experience, lack of funds and time to quantify. Remarkably, most of them agree that it is very important to quantify and use indicators in the different policy cycle phases but they refer to the tendency to replace quantification studies and planning by intuition or local knowledge. Some justify this by the need to act urgently in a very unstable context. On the other hand, it was possible to find some examples of present use of indicators.
Supranational organisations have a role in the demand of, or in the collection of indicators as well as in the set of targets. In Argentina the IMF, the World Bank or the Inter-American Development Bank make local governments use indicators to monitor progress/performance and target areas. Another reason why indicators can be implemented is to control the performance and the implementation of privatised services.
Concern was expressed about the lack of planning tradition with as a consequence less interest in measuring. In any case, there is high recognition by policy makers of the importance of having the best information possible when it comes to take decisions or start implementing a programme or action. Besides, most of the interviewees recognised that they rely not only on quantitative but also on qualitative information and they highly appreciate the combination of both.
In terms of data availability, policy makers recognise the need to work with small areas but they are aware of the difficulties in finding data at such a high level of disaggregation. The economic difficulties that affected the schedule in the implementation of the Census 2001 in Argentina are a clear example of that. According to one interviewee the lack of information does not seem to be a problem but the lack of systematically presented information. Other interviewees considered that quantifying is important but still there is a lack of personnel working on that. Other reactions expressing the same concern:
“The teams we have to perform the tasks are very limited and sometimes they don’t have the technological resources. In our case the staff share one PC and this is very limiting because they take turns to use it.”
“Most of the areas can be approached at a micro scale but for that we require staff”
One interviewee brought the issue that it is not a problem of lack of staff, resources or know how but a problem of “lack of planning culture”:
“There is always a point in the development of the administration that it is necessary to evaluate, to correct and improve. That [evaluation], in general, it is not being incorporated and things are mostly done in an arbitrary and non-rigorous way. That is: it functions essentially based on demand. This is also related to the lack of planning culture”.
It was stated in the interviews that even though they have used indicators they are “still in the search of how to improve the link between indicators and policy decision making” The lack of use of indicators in decision-making was mentioned by other interviewees:
“There is not a practice of using these elements [indicators] when it comes to decision making and in general these decisions are more intuitive or empiric rather than based on information.”
One positive aspect was that the multidimensionality of inequality (and social problems in general) makes policy makers act in cooperation with different areas and secretaries and seem to make them use more indicators. Some of the interviewees also mentioned that due to the need to receive credit from an international agency they started to cooperate with another area within the local government and indicators where presented and exchanged to prepare the reports: “To determine the neighbourhoods [that will receive a credit for urban art] the Inter-American Development Back asked a socio-economic study of these neighbourhoods”.
GIS- based indicators constructed from census data.
After the interviews, it was decided to include the ten most important indicators suggested by the policy makers as a representative set of indicators of aspects of inequality. Income level is not included in the set since there is no income data available but a socio-economic status index is used as a proxy and constructed with factor analysis. Accessibility to internet, although it was considered the lowest important aspect of inequality, is still kept since it is necessary to cover the domain of the accessibility to ICT and new technologies considering its future implications in the quality of life of individuals and that technological diffusion is selective (Castells, 1996). The accessibility to day-care centres (“Centros Crecer”) was included since they represent a policy response to inequalities. Expressed housing needs, calculated from administrative data, are also included. This indicator is also relevant due to the high importance of the habitat in the quality of life of the population and is included as an example of how administrative data can also be used to construct GIS-based indicators.
- Overcrowding; (% of households with 3 or more persons per bedroom) this represents a critical level of housing need.
- Education level; (% of illiterate adults) which represents a group of high vulnerability since urban deprivation is most common among the unskilled. (Langlois and Kitchen, 2001)
- Unemployment; (% of unemployed adults)
- Water Connections; (% of households with tap water inside the house)
- Access to kindergartens and primary schools; (distance from the centroids of the block groups to the nearest kindergarten or primary school represented in a point map)
These five indicators can be considered as the top priorities according to the policy makers and they cover all the inequality domains.
With the use of a desktop GIS (ArcGIS) a selection of 2001 census variables was used to construct and map the five selected indicators. The spatial detail of the census data is determined by the boundaries of each census tract (“fracción censal”). They are established so that the population in each is approximately of 13,500 people, regardless of its geographic size. The National Institute of Statistics and Censuses from Argentina (INDEC) divides each census tract into 15 smaller units called “radio censal” with a population of approximately 900 people. These “radios censales” are built up from individual blocks (“manzanas”) but the data is not available to the public at that level of disaggregation.
Figure 1: Block groups ("radios censales")
As shown in Figure 1, tracts are formed of an average of 15 block groups (“radios censales”). There are 901 Blocks groups in Rosario with and average of 1015 People in each. The following are the resulting GIS-based indicators constructed at block group level. The best off areas are shown in white and the worst off areas in black. A spatial patterning of inequality can be found in the city of Rosario with a concentration of needs in certain areas (See Figures 2 and 3).
Figure 2: Spatial patterning of inequalities in Rosario. Source: own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina)
Figure 3: Accessibility to kindergartens and primary schools (minimum distance from census tract gravity point, in meters) Source: Own analysis
If we analyse the indicators of the best off block group and the worst off block group and between the best 10% and worst 10% it is possible to appreciate the considerable gap between both groups (Table 5).
Aggregation level – Ranking neighbourhood areas
Considering that indicators should be related to policy actions, it is important to see how inequality is depicted at a level of aggregation coincident with the area of intervention of policy actions. In many cities this is the case for the municipality districts. Policy makers might preferably act at district level where neighbourhood participation activities also take place at that level. However, the definition of neighbourhood areas to analyse the data is quite controversial since we come across the problem of agreeing on a common definition of neighbourhood. Not even citizens agree on the boundaries of the neighbourhood. There is no clear boundary acceptance since the different perceived neighbourhoods overlap in space and sometimes the inhabitants even have alternative names and borders.
In the case of Rosario every district is subdivided by the Municipality into six “areas barriales” or neighbourhoods areas. To provide policy makers with an option of ranking those areas, the indicators constructed at block group level were summarised within ArcGis into neighbourhood areas and ranked from the worst off to the best off (See Figure 4 and Table 6)
Figure 4: Neighbourhood areas with at least one census block ranked within the five worst off areas in terms of overcrowding, unemployment, illiteracy rate and tap water inside the house.
GIS-based indicators constructed from administrative data – “Hidden need”
One of the problems of collecting indicators at intra-urban level is the lack of data other than census data –most of the research done on spatial inequality is based on this source-. This poses a conflict to solve, which is to find alternative sources (other than census data). With this in mind, this research starts addressing this issue by including the use of points of expressed demand (Martinez, 2000). Indicators from census data are good to measure indirect need (or demand) but they cannot measure expressed demand coming from the population. While studying the role of GIS in urban planning, Webster (1993) identified expressed demand as a component of descriptive analysis. Webster differentiates between imputed and expressed demand. Imputed demand (or derived demand) is evaluated indirectly by inference from locational information by means of demand indicators; on the other hand, expressed demand is evaluated directly by recording the expressions of demand by members of the public.
In the case of housing needs, for example, needs can be evaluated indirectly by means of indicators. On the other hand, expressed demand/need is evaluated directly by recording the expressions of demand by citizens. The number of registered demands for good housing is also considered a good indicator of the unsatisfied needs and may serve as an alternative indicator (Mega, 1995).
Administrative databases available at the municipalities are very useful to describe inequalities at a low level of detail where census data is not available and as a complimentary data source. In this research data provided by the Public Housing Office (SPV) is used. This database contains data of 12,926 cases of express demand for housing solutions. They can also be aggregated at block or census tract level if needed.
To detect where the expressed demand is coming from, a point map is produced. The data used to elaborate this map was tabular data created at SPV by the Social Work Department. The field selected was that of the address of the person that voluntarily went to the SPV office and asked for a housing solution or expressed his housing problem.
Geocoding and address matching were the GIS processes used to map the expressed demand, where the address provides the locational key of the demand. The address matching capability of ArcGIS can generate point locations that represent events on a map containing (like in this case) a street network ready for geocoding. The map network has address ranges for the right and left sides of the streets assigned to the segments. ArcGIS reads the address from the event tabular data and estimates a map position. After performing the geocoding process, 76% of the addresses were matched.
This approach proved to be efficient to detect cases of housing needs where derived demand via indicators is showing lower levels of demand or hiding the situation at all. This can help in the detection of the “new poor” normally hidden by many indicators . This new poverty is scattered throughout the cities and since they are not living in recognisable poor neighbourhoods they can be found in any middle-class apartment block (Minujin, 1995)
In Figure 5 we can see how points of individuals expressing their housing needs were “hidden” in an area classified as within the 10% best off block groups. At the same time, if we analyse the neighbourhood area (on the left map) we can see how two extreme realities are close to each other: there is a block group (hatched polygon) which belongs to the best 10% areas only 400 m to a block group which belongs to the worst 10% area (grey polygon).
Figure 5 : Neighbourhood areas with contrasting realities. Own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina) and SPV. Expressed demand of housing (each dot represents the demand from one family).
Factor analysis
Finally, a more exploratory approach was chosen to compare the five most important aspects of inequality indicators (selected after the interviews and the literature review) with the factors obtained with factor analysis. Factor analysis derives from a large set of variables, a set of factors that can be thought of as “super variables which represents a cluster of highly correlated census-based variables” (Davies 1984 in: Pacione, 2001). The meaning of each factor can be associated by the original variables with high loadings.
To generate the factors an extended set of variables obtained from the 2001 census was analysed. A further selection of variables was done by discarding those that had low communality and those that generate a “closed system” (e.g. households with cell phone and households without cell phone). A list of 57 variables was finally included to extract the factors using principal components and a rotated (orthogonal) solution (which means that the factors do not correlate with each other but optimally with the common variance in the original variables). From the factor analysis four factors were identified and named according to the variables that contain.
Factor 1: Wealth and Social Status; (shown in the following map)
Factor 2: Family (age);
Factor 3: Family (status - woman headed households);
Factor 4: Housing (inadequacy or housing deprivation).
From this result, we can suggest to consider Factor 1 as a good proxy of income inequality since it is highly related to wealth and social status, with high loadings on variables such as ownership of cell phone, PC with internet connection, cable TV, employment, university level education and private school attendance.
As we concluded in the previous section, if we compare the wealth factor with the points of express demand we can find “hidden” needs in areas classified as wealthy (Figure 6). It is clear that to avoid the problem of ecological fallacy we should not assume that every person living in a wealthy area is wealthy.
Knox and Pinch (2000 p.123) also warn about the dangers of ecological fallacy (i.e. making inferences about individuals with data based on aggregates of people) and put it in this way
“not everyone in a deprived area is necessarily deprived and not every deprived person in an area of ‘multiple deprivation’ is necessarily multiply deprived”.
Figure 6: Own analysis based on 2001 census data provided by INDEC (National Institute of Statistics and Censuses - Argentina) and SPV. Expressed demand of housing (each dot represents the demand from one family).
Conclusions
In 1995, the Municipality of Rosario started a decentralization processes which led to the creation of six districts. They stated the importance of comparing different geographical areas “to adjust equity criteria in the assignment of resources”. As many other local governments around the world, they are faced to the importance of having good information to support policymaking. In this paper, the use of GIS-based indicators was applied in a case study where they prove to be valuable to detect need areas and communicate detailed geographical patterns of inequality. Here it is suggested that GIS has a good opportunity to be integrated into policy making via indicators that are linked to real problems and policy goals (in this case: “to adjust equity criteria in the assignment of resources” or to target need areas).
Some advantages of GIS to construct indicators emerge from this case study. As it was explained in the first section to operationalize indicators, it is necessary to: organize data, to quantify and to communicate. In this case, it was possible to integrate different data sources such as census and administrative data, quantify needs and analyze the gaps between best and worst off areas, and to generate maps to communicate and detect problem areas.
Policy makers confirmed the need to use indicators in a more systematic way and mostly to use them during the problem description phase of policy making. However, many of those interviewed expressed their concerns about issues such as lack of data (or systematized data), lack of staff and/or resources. The lack of a tradition of the use of quantitative analysis within the policy context of Rosario might be improved by the provision of specific information attached to policy goals and specific phenomena with GIS-based indicators. The multidimensional aspect of inequalities and the demand to target need areas with indicators is a good opportunity both to induce the exchange of information between different municipality departments and to implement the use of GIS in policy making.
Beyond this, it is necessary to improve the provision of small area information and a better coordination between municipality departments to generate consistent administrative data that are suitable both for the construction of indicators and for the generation of geographic data.
References
- Bifarello, M. (2000). Public-Third Sector Partnerships: A Major Innovation In Argentinean Social Policy. Fourth Conference of the International Society for Third-Sector Research, Dublin, Ireland.
- Bragos, O., A. Mateos, et al. (2001). Social and Spatial Segregation in Rosario West Side Expansion. 37th International ISOCARP Congress, Utrecht, The Netherlands.
- Broadway, M. J. and G. Jesty (1998). "Are Canadian Inner Cities Becoming More Dissimilar? An Analysis of Urban Deprivation Indicators." Urban Studies 35(9): 1423-1438.
- Castells, M. (1996). The Information Age: Economy, Society and Culture: Vol. I The rise of the network society. Oxford, Blackwell.
- Chakravorty, S. (1996). "A Measurement of Spatial Disparity: The Case of Income Inequality." Urban Studies 33(9): 1671-1686.
- Couclelis, H. (1999). Spatial Information Technologies and Societal Problems. Geographic Information Research: Trans-Atlantic Perspectives. M. Craglia and H. Onsrud. London, Taylor & Francis: 15-24.
- Deelstra, T. and D. Boyd (1997). Indicators for Sustainable Urban Development. Advance Study Course on Indicators for Sustainable Urban Development, Delft, The Netherlands, The International Institute of the Urban Environment.
- Devas, N. and C. Rakodi (1993). Managing fast growing cities. New approaches to urban planning and management in the developing world. London.
- EC (European Communities) (2000). THE URBAN AUDIT. Towards the Benchmarking of Quality of Life in 58 European Cities. Luxemburg.
- European Commission (2000). Towards a Local Sustainability Profile: European Common Indicators. Brussels, Expert Group on the Urban Environment. Working Group on Measuring, Monitoring and Evaluation in Local Sustainability: 12.
- Fundación Banco Municipal (1996). Asentamientos irregulares en Rosario 1996. Rosario, Fundación Banco Municipal.
- Ghose, R. and W. E. Huxhold (2002). "The Role of Multi-scalar GIS-based Indicators Studies in Formulating Neighbourhood Planning Policy." Journal of URISA 14(2): 5 -17.
- Huxhold, W. E. (1991). An introduction to Urban Geographic Information Systems. New York, Oxford University Press.
- Innes, J. E. (1990). Knowledge and public policy: the search for meaningful indicators, New Brunswick, NJ : Transaction Publishers.
- Kingsley Thomas G. (1999). Building and Operating Neighbourhood Indicator Systems, National Neighbourhood Indicators Partnership THE URBAN INSTITUTE.
- Knox, P. and S. Pinch (2000). Urban social geography: an introduction. Harlow, Prentice Hall.
- Kruythoff, H. (1998). Segregation diagnostics, antisegregation remedies, six European case-studies reviewed. Towards undivided cities in Western Europe. New challenges for urban policy. Part 7 Comparative analysis. H. Priemus, R. van Kempen and S. Musterd. Delft, Delft University Press. 7.
- Kunzmann, K. R. (1998). "Planning for spatial equity in Europe." International Planning Studies 3(1): 101-120.
- Langlois, A. and P. Kitchen (2001). "Identifying and Measuring Dimensions of Urban Deprivation in Montreal: An Analysis of the 1996 Census Data." Urban Studies 38(1): 119-139.
- Martinez, J. (2000). "Evaluating housing needs with the use of GIS." Habitat International 24: 501-515.
- Mega, V. (1995). Sustainability indicators for European Cities. European Foundation for the Improvement of Working and Living Conditions. OECD/Rennes Conference on Urban Indicators, Rennes.
- Mega, V. (1996). "Our city, our future: towards sustainable development in European cities." Environment and Urbanisation 8(1): 133-154.
- Minujin, A. (1995). "Squeezed: the middle-class in Latin America." Environment and Urbanization 7(2): 153-165.
- Moor, J. (2000). "Learning Cities, The Global Urban Observatory." Habitat Debate 6(1).
- OECD (1997). Better Understanding Our Cities, The Role of Urban Indicators. Rennes, Organisation for Economic Co-Operation and Development (OECD): 96.
- Pacione, M. (2001). Urban Geography, a global perspective. London, Routledge.
- Plan Estratégico Rosario (PER) (1998). Plan Estratégico Rosario, Diagnóstico y Formulación. Rosario, PER y Agencia Española de Cooperación Internacional.
- Smith, D. M. (1973). The geography of social well-being in the United States: an introduction to territorial social indicators, New York, N.Y., etc.] : McGraw-Hill.
- Smith, D. M. (1994). Geography and social justice. Oxford, Blackwell.
- The World Bank (1996). UMP Working Paper Series. Urban Poverty Research Sourcebook. Module II: Indicators of Urban Poverty.
- UNCHS (1995). Indicators Programme, Volume 1, 2 and 3. Nairobi, UNCHS.
- UNCHS (2000). The Global Urban Observatory's Training Manual: 84.
- UNCHS (2000). URBAN INDICATORS TOOLKIT - GUIDE FOR ISTANBUL +5. Nairobi, UNCHS.
- UNCHS (2001). Cities in a Globalizing World. Global Report on Human Settlements 2001. London, Erthscan.
- UNDP (2000). Chapter 5: Using indicators for human rights accountability. Human Development Report 2000. UNDP. Oxford, Oxford University Press: 89-111.
- UNDP (2001). Human Development Report 2001. Oxford, Oxford University Press.
- United Nations (1992). Rio Declaration On Environment And Development. REPORT OF THE UNITED NATIONS CONFERENCE ON ENVIRONMENT AND DEVELOPMENT. Rio de Janeiro.
- Van Delft, Y. (1997). Indicators for Sustainable Urban Development. Advance Study Course on Indicators for Sustainable Urban Development, Delft, The Netherlands, The International Institute of the Urban Environment.
- Webster, C. J. (1993). "GIS and the scientific inputs to urban planning. Part 1: Description." Environment and Planning B: Planning and Design 20: 709-728.
- Wessel, T. (2000). "Social Polarisation and Socioeconomic Segregation in a Welfare State: The Case of Oslo." Urban Studies 37(11): 1947-1967.