#!  

   "%  

        " ! &  )+%*((,     "   DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

EXECUTIVE SUMMARY

Latin America is characterized by high rates of marginalization and inequality, and the rapid urbanization throughout the region in the last half-century has exacerbated issues of substandard living conditions and security for the poor. In a rapidly changing urban environment, governments and development agencies working in the region need consistent methodologies for locating the most marginalized populations, as well as determining area’s most in need of structural interventions. Poverty alleviation policy can be much more effective when geographically targeted at those in most need.

Mexico has an elaborate government branch working with demographic and geo-spatial data that has developed various systems and variable-indicators for measuring marginalization. This report borrows from such studies and uses Geographic Information Systems (GIS) to create a model for ranking marginalization using a very small unit of analysis for the Metropolitan Area of , located in Nuevo León, . The intention was to provide a scaled-down look at the presence of poverty in a Latin American city that is most often praised for its wealth and advancement in international business. Research was conducted principally by consulting studies from Mexican sources such as Instituto Nacional de Estadística y Geografía (INEGI), Consejo Nacional de la Población (CONAPO), and Instituto Tecnológico de Estudios Superiores de Monterrey (ITESM).

The ultimate motivation for this study is to provide a relatively simple, accessible GIS model for mapping marginalization in Mexican cities. It allows the researcher to locate the most marginalized areas of the population from an analysis of combined variable- indicators and then focus on one policy-related variable (i.e. access to running water) for a specific highly marginalized neighborhood or block. The model is not to be applied without an understanding of the context and the situation of the various marginalization- indicators for the specific region being studied. It can easily be modified and allows for the researcher to select different variables, if appropriate.

2 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

INTRODUCTION

Mapping Marginalization in Mexico

Latin America has been recognized as one of the world’s regions with the highest levels of inequality in relation to the distribution of income, and although the region’s poverty levels may not be as extreme as some others, they remain consistently high. The Economic Commission for Latin America and the Caribbean (ECLAC) has measured that the amount of households in the region living in poverty has hovered between 35 and 40% since the early 1970s (Montes Avilés 2003). The urban areas of Mexico maintain average poverty indices very similar to the overall region; in the year 2000, an average of 10% of Mexican households were measured as indigent while another 37.5% were considered to be poor (Montes Avilés 2003).

As Latin America and the rest of the developing world continue to rapidly urbanize, the dynamics of poverty and security for the marginalized are changing, and governments as well as non-profit agencies are looking for effective programming to combat poverty and ensure safe and sanitary living conditions. In order to maximize the effectiveness of poverty reduction policy, agencies must ensure aid-resources are directed solely at the poor. One way of doing so is to target resources geographically, which requires detailed information on the location of the poor (Fujii 2008). In addition to mitigating aspects of poverty such as basic nutrition and education, comprehensive policy must specifically address the structural rehabilitation of settlements that are the result of rapid and unplanned generation of informal housing.

There is no one way to measure poverty or marginalization; while some methodologies take into account purely economic factors, others give more weight to social indicators. The Consejo Nacional de la Población (CONAPO), Mexican federal agency for demographic and population studies, has developed various methodologies for measuring marginalization with inclusive variables that take into account access to education, living conditions, and income.

The Monterrey Metropolitan Area

The Monterrey Metropolitan Area (MMA) is Mexico’s third largest metropolitan area by population (after Mexico City and ), and is located in the northeastern state of Nuevo Leon. The MMA is officially composed of nine : , Escobedo, Garcia, Guadalupe, Juarez, Monterrey, San Nicolas de los Garza, San Pedro Garcia Garza, and Santa Catarina. The MMA was officially created in the year 1984, and initially included seven municipalities (Garcia and Juarez, located on the outskirts, are the two municipalities that were more recently added). Two additional adjacent municipalities, Santiago and , are beginning to be considered as part of the MMA as well. The core of the urban conurbation is the of Monterrey, which also serves as the capital of state of Nuevo Leon.

 3 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

4 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

The MMA, like other Mexican cities, has experienced an explosion in population since the mid-1900s. In the year 1950, the nine municipalities that today form the MMA had a total population of 339,282, and 87% of these inhabitants were concentrated in the municipality of Monterrey. In the year 2000, the total population of the MMA reached 3,243,466, composing 84.6% of the total state’s population.

Population Growth MMA 1950-2000  



 



 





           

 Source: INEGI

As the urban footprint has expanded, the municipality of Monterrey’s share of the population has decreased. The municipalities that have most recently increased in population are the outer areas of Apodaca, Escobedo, Juarez, and Garcia, all of which approximately doubled their population in the 1990s alone. This indicates a rapid urbanization that transforms municipalities from largely rural areas into the enveloping network of the MMA.

Monterrey has traditionally been a stronghold for manufacturing and industry, likely a contributing factor to its rapid expansion from in-migration. In the year 1950, 46% of the MMA’s economically active population was employed in the manufacturing sector; in the year 2000 that number had decreased to 26.9%. This indicates a rise in other formal sectors, such as finance and business, but also a rise in the informal economic sector brought on by rapid urbanization. While Monterrey is generally considered the best place in all of Latin America to do business, and is home to the wealthiest municipality in the country (San Pedro Garcia Garza), this does not guarantee that the economic benefits are distributed uniformly across the MMA. The Encuesta de Ingresos y Gastos de los Hogares Área Metropolitana de Monterrey (Income and Spending Survey of MMA Households), conducted in 1995, indicated that the total combined income of the lowest- earning 80% of the population of the MMA was less than the combined earnings of the top 10%. When divided into deciles, the top decile earned 27 times the income of the bottom decile. However, Monterrey’s Gini Coefficient (an internationally-applied  5 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 measurement of inequality in which a measurement of 0 indicates total equality while 1 indicates total inequality) is 0.4938, indicating slightly more equality than the country overall, with a Gini Coefficient of 0.5187 (Montes Avilés 2003).

The nine municipalities of the MMA all maintain distinct administration of urban management issues such as transportation, public security, waste management and other services. Article 115 of the Mexican Federal Constitution states that each municipality is to be governed by a directly elected body, without any intermediary governing between the municipal level and the state level. There is even an explicit prohibition of a governing body composed of various municipalities (such as the County level of governance in the United States). Such administrative disjunction leads to inefficiency in all matters of urban management in the MMA. This includes policy approaches to alleviate poverty, from nutritional programs to housing and infrastructure access.

 6 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

PROBLEM STATEMENT & RESEARCH QUESTIONS

The government of Mexico and other similarly structured Latin American countries must use census and survey data to determine where the most marginalized members of its population are located, in order to best allocate funds and programming targeted at poverty alleviation. International development agencies and non-profits would also greatly benefit from a methodology of systematically locating the most marginalized areas within the ‘mega-city’ urban conurbations such as the MMA.

The official Mexican agency for geographic and demographic data, the Instituto Nacional de Estadística y Geografía (INEGI), along with the CONAPO, has various systems for ranking the levels of well being and marginalization for states and municipalities. The state of Nuevo Leon has consistently ranked high in well-being indices in comparison to the remainder of the country. One such index of marginalization study carried out in 1995 took into account four dimensions of marginalization: education, housing, population dispersion, and income. Nuevo Leon was placed in the lowest bracket of marginalization, along with 3 other states and the Federal District (Mexico is composed of 31 states and 1 Federal District). Rankings for marginalization have also been conducted at the municipal level, as indicated by the maps below.

MAP 1.3 Marginalization Ratings for Nuevo León for the year 1995

Monterrey Metropolitan Area

Grade of Marginalization Very low Low Middle High Very High

Source: Consejo Nacional de Población (CONAPO) y Programa de Educación, Salud y Alimentación (PROGRESA), 1998. COMPILED BY CEDEM. Accessed from Montes Avilés, 2003.

 7 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

Such studies can be important first steps for allocating government funds and programming. However, they can also discount the diversity within large geographic areas as well as the sharp contrast of the very poor living ‘next door’ to the very rich, a rampant trend throughout Latin America. For this reason, a consistent methodology must be applied at a low scale of study areas in order to properly identify the most marginalized zones in need of housing and infrastructure interventions.

The most recent indexes calculated represent the average situation for the inhabitants of the municipality. However, just like in all big cities, even if the average of some municipalities does not indicate a high level of marginalization, they can still have towns, neighborhoods, or other areas that are very marginalized. More disaggregated information is required to locate such areas. This occurs in the MMA, where analyzing the municipalities comparatively at the national level doesn’t show high indices of marginalization, but analyzing them individually at a smaller unit, such as AGEBS,1 proves (in addition to the obvious and undeniable presence of an indigent population) that marginalization does exist in the MMA (Montes Aviles 2003, p. 294—translated by author).

This report uses Geographic Information Systems (GIS) to create and test a methodology for identifying and ranking the most marginalized areas within a metropolitan area. The case used here is the MMA, but the hope is that upon refinement the model could be transferred to other municipalities and metropolitan areas within Mexico, and even other Latin American countries. It could provide a useful tool for international groups seeking to rank marginalization at a small scale across larger regions.

In addition to identifying the most marginalized areas within the MMA, the report takes a step further in its analysis by focusing on a specific highly marginalized area to demonstrate how GIS can be used at the block level to identify which residential clusters are in need of specific housing or infrastructural improvements.

This report is a ‘trial study’ in nature, so specific attention will be given to caveats encountered with the data or the analysis/methodology itself. In summary, the main research questions are:

- How can GIS be used to identify and rank the most marginalized areas within a Mexican (or other Latin American) metropolitan area?

- How can GIS be used to pinpoint specific neighborhoods and blocks in need of housing/infrastructure interventions?

- What is the effectiveness of the methodology model and how can it be applied in future studies?

 $ "     #       !   $   8 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

Since the trial study at hand focuses on the MMA, more specific research questions are also proposed:

- Where are the most marginalized areas located within the urbanized area of the MMA?

- Is there a concentration of highly marginalized areas within any one municipality?

- Within areas determined as highly marginalized, where are the neighborhoods and blocks in most need of specific types of structural (housing or infrastructure) intervention?

 9 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

METHODOLOGY

The methodology summary is divided into four sub-sections:

1. Acquire Data 2. Map Marginalization Variables at the AGEB level 3. Generate Ranking of AGEBS based on Marginalization Variables 4. Map Example of ‘Block Level Analysis’ Using Infrastructure Variables

1. Acquire Data

The study required two main types of data: spatial data in the form of GIS shapefiles, and demographic/census data in the form of excel files. Since the project involves mapping Mexican data, and I am based in the US, I was severely limited as to the spatial data I could obtain. While INEGI does provide limited downloadable shapefiles over the Internet, they are not at the level of analysis I needed for the study. Therefore the methodology was in large part determined by the data I could access.

Spatial data were obtained from a disc released to me personally by Dr. Peter Ward, professor at University of -Austin. The disc included shapefiles for the state of Nuevo Leon, specifically a shapefile of all the AGEBS for the state and another for all of the blocks. The AGEBS shapefile did not contain more specific demographic data within its attribute table, while the blocks shapefile did. The original source for this spatial data is INEGI, and the projection & datum are Lambert Conformal Conic, GCS North American 1927.

The demographic data that I used in conjunction with the AGEB shapefiles is from the Censo General de Poblacion y Vivienda 2000 (General Population and Housing Census 2000), conducted by INEGI. The level of detail I required was also not available online, but I was able to obtain it through an INEGI generated software provided to me personally in disc form by a colleague, Cristina Saborio. The software program (SINCE 2000, Sistema de Informacion Censal 2000. Instituto de Informacion Geografica e Informatica INEGI. Aguascalientes, Mexico 2005) is published by INEGI and available for sale in Mexico. I was able to use the software to export .dbf files of demographic data at the AGEB level, which I then opened in excel.

In summary, the data (all from INEGI) used in the analysis were as follows:

o AGEBS shapefiles for Nuevo Leon o Block shapefiles for Nuevo Leon o 2000 Census Data for the AGEBS located in the AMM (the specific tables used were: Employment, Education, Indigenous Language, Migration, Population, and Housing)

 10 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

2. Map Marginalization Variables at the AGEB level - Determine the 10 variables for the marginalization analysis. A similar marginalization study by CONAPO with data from 1990 Census was used as a model to determine the variable indicators for marginalization. - Generate main ‘Variables’ data table. I used Excel to generate a table with the percentages for all 10 variables for each AGEB. - Create ‘AMM_AGEBS’ shapefile. The file I had was for all AGEBs in the state of Nuevo Leon, so I wanted to reduce it to just the AGEBs in the AMM. In ArcMap, I selected all the AGEBs that were in the 9 municipalities of the AMM, and created a layer. I then exported that layer and saved it as ‘AMM_AGEBS.’ - Further Prepare ‘AMM_AGEBS’ shapefile. I joined the ‘Variables’ table to the new ‘AMM_AGEBS’ shapefile, using the code for AGEB as the common field. I then noted that I did not have corresponding Census data for some of the AGEBS in my shapefile. I removed these AGEBS so as to not include them in the analysis, ending with an ‘AMM_AGEBS’ shapefile with a total of 1070 AGEBS. - Choose thresholds for variables. I chose to set a unique threshold for each variable so as to include the top 10% most marginalized AGEBs for that given variable. I calculated these thresholds in Excel. - Run query and create shapefile for each variable. In ArcMap, I added the AMM_AGEBS shapefile. Beginning with the first variable, I ran a query to select all AGEBs greater than or equal to the threshold value. I created a layer of those selected AGEBs and exported it as a shapefile. I repeated for each variable, finishing with ten new shapefiles. - Generate a map for each variable. In ArcMap, I added the AMM_AGEBS shapefile and the first variable shapefile to the same data frame on top of it, and symbolized them appropriately. I repeated for each variable.

3. Generate Ranking of AGEBS based on Marginalization Variables

 Combine variables to generate ‘marginalization ranking.’ From ArcMap, I exported the attribute table for AMM_AGEBS, and then opened it in Excel. Starting with the first variable column, I replaced all values equal to and over the threshold with a 1, and all values under the threshold with a 0. After doing this with all variable columns, I created a new column, ‘GM’ (Grade of Marginalization), the sum of all the variables columns (now made up of 1’s and 0’s) for each AGEB row. The GM column contained values for the ranking ranging from 0 to 9. I saved this excel table as “AMM_AGEBS_grados.”  Generate map to represent ranking of “Marginalized Areas in the AMM.” In ArcMap, I added the AMM_AGEBS shapefile, and joined the

11 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

AMM_AGEBS_grados to it. I symbolized the marginalization grade in three classes, ‘high,’ ‘medium,’ and ‘low.’ The AGEBs with a value of ‘0’ GM I labeled as ‘not included in study,’ since they did not pass the threshold for any of the variables.  Generate Map of Highly Marginalized Areas by Municipality. For this map, I modified the AMM_AGEBS layer by symbolizing the AGEBs based on their Municipality Name field; I made each municipality a faded neutral color without an outline. I also selected all the ‘high’-ranked grade AGEBs and exported them to a shapefile named ‘AGEBS_GMHigh.’ I symbolized ‘AGEBS_GMHigh’ with a cross-hatch fill on top of the Municipalities layer.

4. Map Example of ‘Block Level Analysis’ Using Infrastructure Variables

 Map municipality for block level analysis. At 14, Municipality clearly had the most number of AGEBs ranked with a ‘High’ marginalization. With the AMM_AGEBS shapefile, I selected all AGEBS within General Escobedo and exported them to a new shapefile (‘AGEBs_Escobedo’). I added ‘AGEBs_Escobedo’ to a new dataframe and overlayed the ‘AGEBS_GMHigh’ shapefile, clipping it to ‘AGEBs_Escobedo’ to create the ‘AGEBs_Escobedo_GMHigh’ shapefile. I then added the ‘Manzanas’ (Blocks) shapefile and clipped it to the ‘AGEBs_Escobedo_GMHigh’ shapefile to make the ‘Manzanas_Escobedo_GMHigh’ shapefile. The ‘Manzanas’ shapefile already contained data on water infrastructure in the attribute table. As an example of a specific block analysis for selected AGEBs, I symbolized the principle water infrastructure variable ‘percent of residences with no piped water on the property,’ using 5 classes with Natural Breaks. I highlighted two AGEBs (Study Areas A & B) that I wanted to analyze further using the block-level water infrastructure data.  Generate series of maps of Study Areas A & B to show availability of piped water in the individual blocks. For each study area AGEB, I manually selected the blocks contained within it, created a layer and exported it as a shapefile, i.e. “Manzanas_Area_A.” Then for each study area I made four maps, using the four variables I had data for on water access. I used Natural Breaks with 5 classes to symbolize the percentages of homes for each variable (i.e. “Percentage of Homes without water in the bathroom”).

12 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

FINDINGS

List of Maps

(Series 1. Context Maps—Included in Introduction)

Series 2. Marginalization Variable-Indicator Maps

Map 2.1 People Who Speak an Indigenous Language

Map 2.2 People Who Did Not live in the State 5 years ago

Map 2.3 People Who Didn’t Complete Primaria Level of Schooling

Map 2.4 People Without Any Schooling After the Primaria Level

Map 2.5 People Age 15 and Over Who are Illiterate

Map 2.6 Residences Without Sewage Connection

Map 2.7 Residences With Only One Room

Map 2.8 Residences With Dirt Floor

Map 2.9 People Age 12 and Over Who are Unemployed

Map 2.10 People Age 12 and Over Who Receive 2 Minimum Wages or Less

Series 3. Marginalization Analysis Maps

Map 3.1 Marginalized Areas in the Metropolitan Area of Monterrey

Map 3.2 Highly Marginalized Areas by Municipality

Series 4. Example Block Analysis Maps

Map 4.1 Escobedo Municipality: Block Analysis of Highly Marginalized Areas

Map 4.2 Study Area A: Percent of Homes Without Piped Water Inside the Property

Map 4.3 Study Area A: Percent of Homes Without Piped Water Inside the Home

 13 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

Map 4.4 Study Area A: Percent of Homes Without Bathrooms

Map 4.5 Study Area A: Percent of Homes Without Water Inside the Bathroom

Map 4.6 Study Area B: Percent of Homes Without Piped Water Inside the Property

Map 4.7 Study Area B: Percent of Homes Without Piped Water Inside the Home

Map 4.8 Study Area B: Percent of Homes Without Bathrooms

Map 4.9 Study Area B: Percent of Homes Without Water Inside the Bathroom

14 MAP 2.1 People who speak an indigenous language

1.70% or more of the population age 5 and older Less than 1.70 % of the population age 5 and older

MAP 2.2 People who did not live in the state 5 years ago

11.97% or more of the population age 5 and older Less than 11.97% of the population age 5 and older

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 2.3 People who didn't complete 'primaria' level of schooling

17.74% or more of the population age 15 and older Less than 17.74 % of the population age 15 and older

MAP 2.4 People without any schooling after the 'primaria' level

49.13% or more of the population age 15 and older Less than 49.13% of the population age 15 and older

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 2.5 People age 15 and over who are illiterate

5.89% or more of the population age 15 and older Less than 5.89% of the population age 15 and older

MAP 2.6 Residences without sewage connection

11.84% or more of inhabited residences Less than 11.84% of inhabited residences

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 2.7 Residences with only one room

13.71% or more of inhabited residences Less than 13.71% of inhabited residences

MAP 2.8 Residences with dirt floor

6.17% or more of inhabited residences Less than 6.17% of inhabited residences

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 2.9 People age 12 and over who are unemployed

1.02% or more of the population age 12 and older Less than 1.02% of the population age 12 and older

MAP 2.10 People age 12 and over who receive 2 minimum wages or less

38.46% or more of the population age 12 and older Less than 38.46% of the population age 12 and older

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 3.1 Marginalized Areas in the Monterrey Metropolitan Area

Level of Marginalization Low Medium High

Not Included in Study

Source: INEGI, 2000 General Census 02.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 3.2 Highly Marginalized Areas by Municipality

General Garcia Escobedo

Apodaca

San Nicolas de los Garza

Monterrey Santa Guadalupe Catarina

San Pedro Juarez Garcia Garza

Number of AGEBs Municipality ranked 'Highly Marginalized' Apodaca 0 Garcia 0 General Escobedo 14 Guadalupe 5 Juarez 1 High Marginalization Monterrey 8 (by AGEB*) San Nicolas de los Garza 0 San Pedro Garcia Garza 0 Santa Catarina 3 AGEB= "Area Geoestadistica Basica"

Source: INEGI, 2000 General Census 0 2.5 5 10 15 20 Compiled by Dana Stovall on 12.5.08 Kilometers F MAP 4.1 General Escobedo Municipality:

Block Analysis of Highly Marginalized Areas

Study Area B

Study Area A

Percent of homes with no piped water on the property 0% - 15% 16% - 40% 41% - 65% 66% - 90% 91% - 100%

Source: INEGI, 2000 General Census 0 1.25 2.5 5 7.5 10 Compiled by Dana Stovall on 12.11.08 Kilometers F General Escobedo MAP 4.2 Municipality Study Area A

Percent of homes without piped water inside the property

0% - 5% 6% - 15% 16% - 30% 31% - 50% 51% - 85%

MAP 4.3 Study Area A

Percent of homes without piped water inside the home

11% - 40% 41% - 60% 61% - 75% 76% - 85% 86% - 100%

Source: INEGI, 2000 General Census 075 150 300 450 600 Compiled by Dana Stovall on 12.11.08 Meters F General Escobedo MAP 4.4 Municipality Study Area A

Percent of homes without bathrooms

0% 1% - 5% 6% - 10% 11% - 12% 13% - 16%

MAP 4.5 Study Area A

Percent of homes without water inside the bathroom

50% - 70% 71% - 80% 81% - 90% 91% - 95% 96% - 100%

Source: INEGI, 2000 General Census 075 150 300 450 600 Compiled by Dana Stovall on 12.11.08 Meters F MAP 4.6 Study Area B

Percent of homes without piped water inside the property

92% - 100% General Escobedo Municipality

MAP 4.7 Study Area B

Percent of homes without piped water inside the home

92% - 100%

Source: INEGI, 2000 General Census 075 150 300 450 600 Compiled by Dana Stovall on 12.11.08 Meters F MAP 4.8 Study Area B

Percent of homes without bathrooms

0% 1% - 5% 6% - 8% 9% - 11% General Escobedo Municipality 12% - 15%

MAP 4.9 Study Area B

Percent of homes without water inside the bathroom

100%

Source: INEGI, 2000 General Census 075 150 300 450 600 Compiled by Dana Stovall on 12.11.08 Meters F DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

ANALYSIS & CONCLUSIONS

Specific interpretation and analysis of the maps is organized below by series. Each series may also include a section of additional general comments.

Series 2. Marginalization Variable-Indicator Maps

Map 2.1 ‘People Who Speak an Indigenous Language’ and Map 2.2 ‘People Who Did Not live in the State 5 years ago’ both represent those who migrated to the MMA. Both show a large concentration in the southwestern portion of the urban footprint, where the Municipality of San Pedro Garcia Garza is located. San Pedro Garcia Garza is not only the wealthiest municipality in the MMA, but the entire country. The concentration of individuals who are recent migrants and/or indigenous language speakers is a direct indication of those who work in domestic service in this the wealthiest zone of the city. Many young women who work as maids and caregivers in homes are from a rural area; in addition, it is custom that they reside in the home where they work, and they are included in the Census count for that home. These variables are therefore an important indication of the premise that a highly marginalized sector can and does exist in even those municipalities that have received the lowest ratings of marginalization. The migration variables provide important information for the provision of some types of social services, but are less important when focusing specifically on structural settlement improvements (since most of the population represented here is assumed to be living in the home in which they work).

Map 2.3 ‘People Who Didn’t Complete Primaria Level of Schooling,’ Map 2.4 ‘People Without Any Schooling After the Primaria Level,’ and Map 2.5 ‘People Age 15 and Over Who are Illiterate’ represent the educational variables selected to indicate marginalization. All three variables mapped in a very similar pattern, with many of the same AGEBs being represented in all three. There is a concentration of AGEBs on the outskirts of the urban footprint and very few AGEBs in the central core area.

Map 2.6 ‘Residences Without Sewage Connection,’ Map 2.7 ‘Residences With Only One Room,’ and Map 2.8 ‘Residences With Dirt Floor’ are the variables indicating substandard living conditions. As expected, they all follow a similar pattern. There is a heavy concentration for all variables in the northwestern portion of the city. There is one AGEB in the central region of the MMA that is mapped for all three variables (situated just northeast of the downtown core). This is unusual, since any significant concentration of substandard housing is expected to be located on the fringes. The particular AGEB represented here could be a unique case of a recent informal settlement on an abandoned territory near the core; or, it could simply be a data error. Any program targeted at housing improvements would be advised to investigate further as to the situation in this particular zone. Lastly, San Pedro Garcia Garza is the one municipality that does not house any AGEBs that qualified under the housing variables.

 27 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

Map 2.9 ‘People Age 12 and Over Who are Unemployed’ and Map 2.10 ‘People Age 12 and Over Who Receive 2 Minimum Wages or Less,’ the employment variable maps, seem to be the most sporadic. A main reason for this is the nature of the variables themselves. ‘People Age 12 and Over Who are Unemployed’ is meant to represent those who are economically active and find themselves currently unemployed. The 2000 Census has a section for all members of the household who are 12 and older, and this is where the questions about employment are asked. The inherent fault here is that many household members over the age of 12 may not even be economically active (i.e. students, the elderly). The variable ‘People Age 12 and Over Who Receive 2 Minimum Wages or Less’ (considered the ‘popular’ income level, even below those earning at the ‘low’ income level) also falls under the same caveat. In addition, it only indicates individual income, and negates household factors. A single person household earning 2 minimum wages is less marginalized than a single mother of three earning 2 minimum wages, and perhaps more marginalized than a household where four members are earning 2 minimum wages.

The thresholds for each variable were selected in order to maintain consistency by including only the ‘worst off’ 10% of each variable. This meant that some variables (such as ‘People who speak an indigenous language’) had extremely low thresholds because the overall presence of that variable is extremely low. Even if some such variables ‘appear’ to be inconsequential, their inclusion is important when creating a model that can be applied in other areas of the country and used comparatively. For example, the MMA may have a very low percentage of inhabitants who speak an indigenous language simply because of its geographic location. Another city further to the south would likely generate a higher threshold for the indigenous language variable.

In summary, all of the variables here provide their own caveats. The selection of variables was based on similar marginalization studies conducted by the Consejo Estatal de Población de Nuevo León. With the vast options of variables available through the 2000 Census data, I determined that borrowing the variables used by a government conducted study was the best manner to maintain consistency. I would conclude that the 10 variables used in this study are appropriate indicators of general marginalization, with the exception of the income/employment variables, which would be better suited if replaced with more accurate data, perhaps from a survey outside the 2000 Census.

The model generated here may also be applied using completely different variables. While this study wanted to first determine the most marginalized areas of the city based on a cross-section of educational, income, origin, and housing variables, the researcher can easily conduct the analysis with a more specific set of variables geared toward one issue, such as housing conditions.

Series 3. Marginalization Analysis Maps

Map 3.1‘Marginalized Areas in the Metropolitan Area of Monterrey’ shows the ranking (high, medium, or low) for all AGEBs that passed the threshold for at least one variable

 28 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

indicator. “High” are those AGEBs with marginalization values 7, 8, or 9; “Medium” are those with values of 4, 5, or 6; and “Low” are those with values 1, 2, and 3. AGEBs that did not pass the threshold for any variable indicators (a value of 0) are labeled ‘Not Included in Study’ because they were not included in the ranking analysis. As expected, the areas with highest rankings of marginalization are located on the fringes of the urban footprint. There are a total of 31 ranked “High,” 69 ranked “Medium” and 310 “Low.”

Map 3.2 ‘Highly Marginalized Areas by Municipality’ shows the highly marginalized areas in relation to the municipalities. In the Mexican case, this map is an essential step in the analysis of marginalization across the metropolitan area. As outlined in the introduction, the 9 municipalities of the MMA maintain separate administration and programming. It is very important to note that the AGEBs included in the analysis compose the urban footprint of the MMA. The recently added municipalities of Garcia and Juarez contain additional AGEBs that are still considered ‘rural’ and are located outside of the immediate reach of the MMA. Therefore, it is important to note that although Map 3.2 marks the municipalities it is not inclusive of their reach. The map is also limited because I removed a few AGEBs from the outer areas for which I did not have 2000 Census data. Therefore, although Garcia and Juarez are by many indices the poorest municipalities in the MMA, they are not represented as such by this map due to their limited inclusion in the analysis.

Series 4. Example Block Analysis Maps

Map 4.1 ‘Escobedo Municipality: Block Analysis of Highly Marginalized Areas,’ is a close-up map of one municipality and its 14 ‘highly marginalized’ AGEBs. I selected Escobedo because it clearly had the largest amount of AGEBS ranked ‘highly marginalized.’ The purpose of the Series 4 maps is to demonstrate how a smaller-scale analysis can be conducted within the highly marginalized areas to determine which blocks are in need of a particular service. For example, a social service agency may want to map data on household nutrition, or the government may want to map where the highest concentration of homes with dirt floors are located. Since I wanted to focus on structural improvements, I chose the water access variables that I had for the block level. Map 4.1 shows an example of how the principle water access variable (residences with no piped water on the property) can be mapped across all the highly marginalized AGEBs for one municipality.

I used Map 4.1 to select two AGEBs to serve as example study areas (‘A’ and ‘B’) for an even closer-up analysis. Initially, I wanted to select an AGEB that had a ranking of 9, the highest output in the analysis. However, I found that the few AGEBs with a ranking of 9 had too little available data at the block level. Some were AGEBs located along a river or mountain range that had very few delineated streets and blocks with household data. So, I selected two AGEBs that showed prominent block formations and that contained sufficient data and a substantial number of houses.

 29 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

Maps 4.2-4.9 show how the water access variables can be mapped for a single AGEB. These maps could be used by a municipal or private entity that is interested in water infrastructure improvements for the most marginalized areas of the municipality. The ‘Study Area A’ maps show a diversity of access to piped-in water within one neighborhood. In contrast, ‘Study Area B’ is an example of a neighborhood in which the clear majority of all the blocks has no access to piped water. This is likely an area where residents are receiving all of their water from a municipal distribution truck or from a publically located tap. The ability to locate such areas remotely using GIS could be very helpful to development agencies.

The purpose of these maps was to indicate how one variable could be analyzed at the block level. Any of a number of variable types could be mapped based on the researcher’s interest.

 30 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

REFERENCES

Calderón Rendón, Gaby, Elvira Naranjo Priego & Gabriela Siller Pagaza (2003). Demografía. UN DIAGNÓSTICO PARA EL DESARROLLO, Volumen 1, pp. 243-269. Instituto Tecnológico y de Estudios Superiores de Monterrey. Monterrey N.L. México.

Fujii, Tomoki (2008). How Well Can We Target Aid with Rapidly Collected Data? Empirical Results for Poverty Mapping from Cambodia. World Development, 36 (10), 1830–1842.

Guajardo Alatorre, Alicia Angélica, (2003). Análisis Estratégico del Área Metropolitana de Monterrey. UN DIAGNÓSTICO PARA EL DESARROLLO, Volumen 1, pp. 460-498. Instituto Tecnológico y de Estudios Superiores de Monterrey. Monterrey N.L. México.

Montes Avilés, Verónica (2003). Condición Socioeconómica. UN DIAGNÓSTICO PARA EL DESARROLLO, Volumen 1, pp. 271-312. Instituto Tecnológico y de Estudios Superiores de Monterrey. Monterrey N.L. México.

 31 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008 

APPENDIX

Step-by-Step Methodology The research led me to a study conducted with 1990 data that served as my model. Principally, I borrowed the same variables that this study used. The map generated with the 1990 data is below, followed by my methodology steps.

Levels of Marginalization in the MMA for the year 1990

Level of Marginalization, 1990 (by AGEB)

Low

Medium

High

Very High

Not Specified

SOURCE: Consejo Estatal de Población de Nuevo León (COESPO), 1993. COMPILED BY CEDEM. Accessed from Montes Avilés, 2003.

 Determine 10 variables for the analysis. The ‘model study’ of marginalization (see above) was conducted with 1990 data. The 10 variables used were as follows:

 32 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

1. Persons age 5 and older who speak an indigenous language. 2. Persons age 5 and older who did not reside in the state 5 years ago. 3. Persons age 15 and older who did not complete ‘primaria’ level of school. 4. Persons age 15 and older without any schooling post-‘primaria.’ 5. Persons age 15 and older who are illiterate. 6. Residences without sewage connection. 7. Residences with only one room. 8. Residences with dirt floors. 9. Economically active persons who are currently unemployed. 10. Households that receive 2 or less minimum salaries.

In order to carry out a similar study with the 2000 Census data, I had to choose the 10 variables that were closest as possible to the 1990 data variables. The 10 variables from the 2000 Census I chose were:

1. Persons age 5 and older who speak an indigenous language (HLENIND). 2. Persons age 5 and older who did not reside in the state 5 years ago (PNORESENT95). 3. Persons age 15 and older who did not complete ‘primaria’ level of school (PRIMIN). 4. Persons age 15 and older without any schooling post-‘primaria’ (P15MSINPOS). 5. Persons age 15 and older who are illiterate (PANALF15YM). 6. Residences without sewage connection (VIPASDRE). 7. Residences with only one room (VIPASCUARED). 8. Residences with dirt floors (VIPASPITI). 9. Persons age 12 and older who are currently unemployed (PODESOC). 10. Persons age 12 and older who receive 2 or less minimum salaries (INGMM1A2SM).

Note that variables #9 & # 10 are distinct than variables #9 & # 10 in the 1990 study. This is likely due to a modification in the census.

 Make main “Variables” data table with percentages of all 10 variables for each spatial unit (AGEB). The data for the 10 variables were located in numerous excel tables so I had to cut and paste all necessary 10 fields into a new table, organized by the AGEB codes. For each variable, I used additional Census data to calculate a percentage. I then placed all 10 variables in percentage form (organized by AGEB code) into my final excel table. This is the ‘Variables’ table.

 Prepare main ‘AMM_AGEBS” shapefile. The file I had was for all AGEBs in the state of Nuevo León, so I wanted to reduce it to just the AGEBs in the MMA. I did this by using the ‘Select by Attribute’ function. I selected using the field containing the name of the municipality where each AGEB is located, and entered

 33 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

the names of the 9 MMA municipalities. Once I had the appropriate AGEBs selected I created a layer from the selection and then exported the layer to make a shapefile. I named this shapefile ‘AMM_AGEBS’. I added this ‘AMM_AGEBS’ shapefile to a new map and also added the newly created ‘Variables’ data table. I joined ‘Variables’ to ‘AMM_AGEBS,’ using the AGEB code as the common field. I then opened the attribute table and noted that some AGEBs included in my shapefile did not have corresponding data from the ‘Variables’ table. I decided to take out those outlying AGEBS since I didn’t have Census data for them. I also removed some of the AGEBS in the municipality of Garcia and some of the AGEBS in Juarez. I ‘removed’ these AGEBS by starting an editing session in the Editing Toolbar, and deleting the fields from the attribute table. In the end, I had an “AMM_AGEBS shapefile” with a total of 1070 AGEBS.

 Choose thresholds for the variables. I had to familiarize myself with the data in order to choose a threshold for what AGEBs would be included in the marginalization analysis. In an excel table, I set all the variables for the 1070 AGEBs to display in descending order. I then navigated to line 107 (the top 10% threshold for the variables) and noted the values at the line. They were as follows: 1. HLENIND—1.70% 2. PNORESENT95—11.97% 3. PANALF15YM—5.89% 4. PRIMIN –17.74% 5. P15MSINPOS—49.13% 6. VIPAPITI—6.17% 7. VIPACUARED—13.71% 8. VIPASDRE –11.84% 9. PODESOC –1.02% 10. INGMM1A2SM—38.46%

By setting the above values as the threshold, I was including in the analysis the top 10% most marginalized for each variable.

 Run query and create layer for each variable. In ArcMap, I added the AMM_AGEBS shapefile. Starting with HLENIND, I used the ‘Select by Attribute’ function to run a query. I entered “HLENIND” “greater than or equal to” “1.70.” (Note: 1.70 is the threshold value for HLENIND as indicated above.) Once I had the selection (of all AGEBs with a HLENIND value greater than or equal to 1.70), I created a layer from that selection. Once I had the new layer, I exported it as a shapefile to my data folder. I gave this first layer the name ‘HLENIND,’ for the variable it indicated. I added the required map elements

 34 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

(title, north arrow, scale, source, legend, author, and date). I then repeated these steps with the remaining nine variables, finishing with ten new shapefiles.

 Generate a map for each variable. I added the AMM_AGEBS shapefile to a new dataframe and then added the new HLENIND shapefile on top of it. I symbolized the HLENDIND layer with a bright color and the AMM_AGEBS layer with a neutral gray to serve as a background. I did this for each variable in it’s own separate data frame. I formatted a layout with two dataframes on one sheet, in order to end with 5 pages. I made sure to create the first sheet with the desired layout, and then used it as a template for the remaining.

 Combine variables to generate ‘marginalization ranking.’ From ArcMap, I opened the attribute table and choose the Options button. I chose the option to export the attribute table for AMM_AGEBS as a .dbf file, and then I opened it in Excel. I deleted all of the rows of AGEB variables that I did not need for my analysis. Starting with the first marginalization variable column (HLENIND), I sorted the column in descending order. I then proceeded to replace all values equal to and over the threshold with a 1, and all values under the threshold with a 0. I repeated this for each of the 9 remaining variable columns. Then I created a new column and titled it ‘GM’ (Grade of Marginalization), and entered in it the sum of all the variables columns (now made up of 1’s and 0’s) for each AGEB row. The GM column contained values for the ranking ranging from 0 to 9. I saved this excel table as “AMM_AGEBS_grados.”

 Generate map to represent ranking of “Marginalized Areas in the AMM.” In ArcMap, I added the AMM_AGEBS shapefile, and then added the AMM_AGEB_grados Excel table; then I joined the two together. I wanted to symbolize the newly created ‘GM’ column. The column contained values from 0 to 9 and I wanted to group them into my own categories (low, medium, and high). In the symbology tab I selected ‘Quantities’ and selected “AMM AGEBS GM” for the value field and ‘none’ for the normalization field. In the Classification box, I set 4 classes and chose the ‘Manual’ breaks method. This allowed me to set the breaks to 0, 3, 6, and 9. I then chose colors for each class and changed the labels for each (0=Not included in study; 0.000001-3= Low; 3.000001-6= Medium; 6.000001-9= High.) I added the required map elements.

 Generate Map of Highly Marginalized Areas by Municipality. I used the previous map as a template to begin this map. First I used the ‘Select by Attribute’ tool and entered: “AMM_AGEBS_GM=7 OR AMM_AGEBS_GM=8 OR AMM_AGEBS_GM=9;” this gave me a selection of all the AGEBs I determined to have a ‘high’ grade of marginalization. I created a layer of the  35 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

selection and exported it to create the shapefile ‘AGEBS_GMHigh.’ I added the new shapfile to the map. I adjusted the symbology for the AMM_AGEBS layer; in ‘Categories’ I selected the field for the names of the municipality where each AGEB is located and I then selected neutral fill colors with no outline for each Municipality. In the Display tab, I adjusted the transparency of the layer to fade the colors out a bit. In the layout, I added text labels over each Municipality. Finally, I symbolized the ‘AGEBS_GMHigh’ layer with a narrow crosshatch over the municipalities, to ensure the reader could see both layers. Finally I studied the attribute table of ‘AGEBS_GMHigh’ to determine how many highly marginalized AGEBs were located in each municipality. I added this information to a table in Excel and placed it in the map layout.

 Map one municipality for block-level analysis. Based on the previous map, I determined the municipality with the highest number of highly marginalized AGEBs (it was General Escobedo, at 14). I started a new ArcMap document and added the AMM_AGEBs shapefile in a new dataframe. I used the ‘Select by Attributes’ tool to select all AGEBs located in Escobedo, and I then made a layer out of them and exported them to a shapefile, ‘AGEBS_Escobedo.’ I added it to the dataframe and removed AMM_AGEBS. I also added the ‘AGEBS_GMHigh’ shapefile. I used the Clip tool to clip ‘AGEBS_GMHigh’ to ‘AGEBS_Escobedo,’ creating the ‘AGEBs_Escobedo_GMHigh’ shapefile. I then added the ‘Manzanas’ shapefile, which contained the block shapefiles for the entire state. I clipped it to the ‘AGEBs_Escobedo’ layer to create the Manzanas_Escobedo’ shapefile. I clipped Manzanas_Escobedo’ to ‘AGEBs_Escobedo_GMHigh’ make another layer, Manzanas_Escobedo_GMHigh.’ I removed the ‘AGEBs_Escobedo’ layer and symbolized the Manzanas_Escobedo’ layer with a faded gray color. I made sure ‘Manzanas_Escobedo_GMHigh’ was on top and I symbolized it as a ‘Quantity’ using data already contained in the attribute table. I entered as the value the variable ‘homes with no piped water on the property’ and entered as the normalization ‘number of homes.’ This allowed me to symbolize ‘percent of homes with no piped water on the property.’ I used 5 classes with Natural Breaks. I then examined the attribute table for ‘Manzanas_Escobedo_GMHigh’ to select two AGEBs with sufficient homes and population to provide good examples of block level analysis. I used graphic elements to highlight these areas on the map, labeling them ‘Study Area A’ and ‘Study Area B.’ I added the appropriate map elements.  Generate series of maps of Study Areas A & B to show availability of piped water in the individual blocks. In a new dataframe, I added the ‘Manzanas_Escobedo_GMHigh’ shapefile. I manually selected the blocks that composed the AGEB I had chosen for ‘Study Area A.’ I created a layer from the

 36 DANA STOVALL – LOCATING THE MARGINALIZED IN MEXICAN CITIES – CRP 386, FALL 2008

selection and exported it to the shapefile ‘Manzanas_Area_A.’ I had four variables concerning water access that I wanted to represent with this shapefile, so I created four small maps, borrowing the layout from my 10 variable maps. The first variable was ‘homes with no piped water on the property,’ and I normalized it by ‘number of homes’ to get percentage values. I followed the same steps in new data frames for the remaining 3 variables ‘homes with no piped water in the house;’ ‘homes with no bathroom;’ ‘homes with no piped water in the bathroom.’ I normalized all values by number of homes, and used Natural Breaks with 5 classes when possible (some had too few differentiation). I used a different color gradation for each variable since the breaks were different. Upon completing the 4 maps for Study Area A, I used them as templates and following the same steps for Study Area B. I made sure to include all necessary map elements, finishing with 8 block-level analysis maps.

 37