### Table 5 Decomposition of inequality, by regional subgroup (GE0) Level of decomposition Number of subgroups Within-group (%) Between-group (%) Ecuador

2003

"... In PAGE 21: ... Ratios of the respective components with the overall inequality level provide a measure of the percentage contribution of between-group and within-group inequality to total inequality. In Table5 we examine how within-group inequality evolves at progressively lower levels of regional disaggregation in our three countries. At one extreme, when a country-level perspective is taken, all inequality is, by definition, within-group.... In PAGE 23: ...The same pattern obtains in Madagascar and Mozambique ( Table5 ). In all three countries no less than three quarters of all inequality is attributable to within-community differences, even after one has disaggregated down to a very low level (corresponding, in our countries, to the lowest level of central government administration).... ..."

### Table 4 Decomposition of Inequality by Regional Sub-Group (GE0)

2003

"... In PAGE 16: ... Ratios of the respective components with the overall inequality level provide a measure of the percentage contribution of between-group and within-group inequality to total inequality. In Table4 we examine how within-group inequality evolves at progressively lower levels of regional disaggregation in our three countries. At one extreme, when a country- level perspective is taken, all inequality is, by definition, within-group.... In PAGE 17: ... In urban areas of Ecuador, the within-group share, across 664 urban zonas (with 1300 households on average) is only slightly lower at 77%. The same pattern obtains in Madagascar and Mozambique ( Table4 ). In all three countries no less than three quarters of all inequality is attributable to within-community differences, even after one has disaggregated down to a very low level (corresponding, in our countries, to the lowest level of central government administration).... ..."

Cited by 2

### Table 7: Regional Inequality in Illiteracy Rate and Infant Mortality Rate (IMR)

2003

"... In PAGE 10: ... Moreover, it appears that the gender gap has increased between 1990 and 1995. Table7 further displays the spread in the illiteracy rate across rural and urban areas, with the Gini and Generalized Entropy (GE) as inequality measures. The GE 2 The data in 1964, 1981 and 1990 are from the census, while the information in 1995 is from a one percent population survey.... In PAGE 11: ... Inequality is calculated using the population weighted values of illiteracy for spatial units at the highest level of disaggregation for which data is available. In the top panel of Table7 , the first two columns show that the Gini and the GE at the national level declined from 1964 to 1981 and then increased from 1981 to 1995. The same pattern holds true for inequalities across rural areas, as shown in the third column for the GE measure.... In PAGE 11: ... As is well known, the GE family of inequality measures can be decomposed into the sum of a within and a between group component, for any given partitioning of the population into mutually exclusive and exhaustive groups. The fifth and sixth columns of Table7 present the evolution of the within and the between group components of inequality. Both components rose in the post-reform period.... In PAGE 11: ...nequality. Both components rose in the post-reform period. Using the within-inequality and between-inequality, we can calculate the polarization index following the method outlined by Zhang and Kanbur (2001).4 As shown in the last column in Table7 , rural and urban areas became increasingly polarized from 1981 to 1995. The above inequality analysis, based on more disaggregated data, offers a snapshot for each of four years.... In PAGE 12: ...and district level as shown in Table7 . The rural regional income inequality, measured by the Gini coefficient, increased by from 13.... In PAGE 12: ... Using the data set, we can further examine the regional distribution of IMR. As shown in the lower panel of Table7 , overall regional inequality increased from 1981 to ... ..."

### Table 3: Measures of inequality for 1996 and 2000 by region. Measures of inequality

2003

"... In PAGE 10: ... The national level of expenditure inequality, measured by the Gini coefficient, reached 0.378 in 1999/2000 ( Table3 ). Urban areas of Egypt had higher levels of inequality in comparison with rural areas.... In PAGE 11: ... 11 capita expenditure, measured by Gini coefficient ( Table3 ), increased from 0.345 in 1995/96 to 0.... In PAGE 11: ...rom .374 to .396 in the Metropolitan region). At the same time, the Gini coefficients declined, by about three percentage points, for both urban and rural areas of Lower Egypt ( Table3 ). Using the Theil index (Theil 1979) to decompose the inequality at the national ... ..."

Cited by 1

### Table 6a Correlates of mean Log deviation (GE0) in rural Ecuador Parroquia-level regression (915 parroquias) Basic regression + expenditure

2003

"... In PAGE 28: ... We ask whether inequality levels are correlated with location, controlling for both demographic characteristics of the communities (population size and demographic composition), and mean per capita consumption. Table6 a for rural Ecuador finds strong evidence that inequality in the parroquias of the eastern, Oriente, region is significantly higher than the province of Pichincha in the central, mountainous, Sierra, region. Communities located in provinces in the western, coastal, Costa, region tend to be more equal, significantly so in the provinces of Manabi, Los Rios, Guayas, and El Oro.... In PAGE 30: ... Possibly there exists greater heterogeneity in household size in those parroquias with large population shares of young children and that this translates into greater inequality of per capita consumption. In urban Ecuador ( Table6 b) the relatively low inequality in the Costa region is again observed. Relative to the zonas in the capital Quito, inequality in all zonas of the Costa region tends to be significantly lower.... In PAGE 31: ... Table6 b Correlates of mean Log deviation (GE0) in urban Ecuador Zona-level regression (660 zonas) Basic regression + expenditure Log population -0.013 (0.... In PAGE 32: ... Table6 c Correlates of mean Log deviation (GE0) in rural Madagascar Firaisana-level regression (1,117 firaisanas) Basic regression + expenditure Log population 0.010 (0.... In PAGE 32: ...55 when all variables are included. In urban Madagascar, the explanatory power is even greater ( Table6 d). Here, unlike rural areas, population size is significantly negatively associated with inequality.... In PAGE 33: ... Table6 d Correlates of mean Log deviation (GE0) in urban Madagascar Firaisana-level regression (131 firaisanas) Basic regression + expenditure Log population -0.014 (0.... In PAGE 33: ...ercent. Excluded groups are Antsiranana and and population age 15-49. (the capital province), Fianarantsoa, and Toliara more unequal than the urban areas in the rest of the country. Table6 e confirms that in Mozambique, too, geographic variables are key indicators of local-level inequality, controlling for population characteristics, mean expenditure levels, and urban/rural differences. Compared with Maputo City, the rest of the country has significantly less inequality.... In PAGE 34: ... Table6 e Correlates of mean Log deviation (GE0) in Mozambique Administrative post-level regression (464 administrative posts) Basic regression + expenditure + urban Percent aged 0-5 -0.002 (0.... ..."

### Table 6c Correlates of mean Log deviation (GE0) in rural Madagascar Firaisana-level regression (1,117 firaisanas) Basic regression + expenditure

2003

"... In PAGE 28: ... We ask whether inequality levels are correlated with location, controlling for both demographic characteristics of the communities (population size and demographic composition), and mean per capita consumption. Table6 a for rural Ecuador finds strong evidence that inequality in the parroquias of the eastern, Oriente, region is significantly higher than the province of Pichincha in the central, mountainous, Sierra, region. Communities located in provinces in the western, coastal, Costa, region tend to be more equal, significantly so in the provinces of Manabi, Los Rios, Guayas, and El Oro.... In PAGE 29: ... Table6 a Correlates of mean Log deviation (GE0) in rural Ecuador Parroquia-level regression (915 parroquias) Basic regression + expenditure Log population 0.0169 (0.... In PAGE 30: ... Possibly there exists greater heterogeneity in household size in those parroquias with large population shares of young children and that this translates into greater inequality of per capita consumption. In urban Ecuador ( Table6 b) the relatively low inequality in the Costa region is again observed. Relative to the zonas in the capital Quito, inequality in all zonas of the Costa region tends to be significantly lower.... In PAGE 31: ... Table6 b Correlates of mean Log deviation (GE0) in urban Ecuador Zona-level regression (660 zonas) Basic regression + expenditure Log population -0.013 (0.... In PAGE 32: ...55 when all variables are included. In urban Madagascar, the explanatory power is even greater ( Table6 d). Here, unlike rural areas, population size is significantly negatively associated with inequality.... In PAGE 33: ... Table6 d Correlates of mean Log deviation (GE0) in urban Madagascar Firaisana-level regression (131 firaisanas) Basic regression + expenditure Log population -0.014 (0.... In PAGE 33: ...ercent. Excluded groups are Antsiranana and and population age 15-49. (the capital province), Fianarantsoa, and Toliara more unequal than the urban areas in the rest of the country. Table6 e confirms that in Mozambique, too, geographic variables are key indicators of local-level inequality, controlling for population characteristics, mean expenditure levels, and urban/rural differences. Compared with Maputo City, the rest of the country has significantly less inequality.... In PAGE 34: ... Table6 e Correlates of mean Log deviation (GE0) in Mozambique Administrative post-level regression (464 administrative posts) Basic regression + expenditure + urban Percent aged 0-5 -0.002 (0.... ..."

### Table 6b Correlates of mean Log deviation (GE0) in urban Ecuador Zona-level regression (660 zonas) Basic regression + expenditure

2003

"... In PAGE 28: ... We ask whether inequality levels are correlated with location, controlling for both demographic characteristics of the communities (population size and demographic composition), and mean per capita consumption. Table6 a for rural Ecuador finds strong evidence that inequality in the parroquias of the eastern, Oriente, region is significantly higher than the province of Pichincha in the central, mountainous, Sierra, region. Communities located in provinces in the western, coastal, Costa, region tend to be more equal, significantly so in the provinces of Manabi, Los Rios, Guayas, and El Oro.... In PAGE 29: ... Table6 a Correlates of mean Log deviation (GE0) in rural Ecuador Parroquia-level regression (915 parroquias) Basic regression + expenditure Log population 0.0169 (0.... In PAGE 30: ... Possibly there exists greater heterogeneity in household size in those parroquias with large population shares of young children and that this translates into greater inequality of per capita consumption. In urban Ecuador ( Table6 b) the relatively low inequality in the Costa region is again observed. Relative to the zonas in the capital Quito, inequality in all zonas of the Costa region tends to be significantly lower.... In PAGE 32: ... Table6 c Correlates of mean Log deviation (GE0) in rural Madagascar Firaisana-level regression (1,117 firaisanas) Basic regression + expenditure Log population 0.010 (0.... In PAGE 32: ...55 when all variables are included. In urban Madagascar, the explanatory power is even greater ( Table6 d). Here, unlike rural areas, population size is significantly negatively associated with inequality.... In PAGE 33: ... Table6 d Correlates of mean Log deviation (GE0) in urban Madagascar Firaisana-level regression (131 firaisanas) Basic regression + expenditure Log population -0.014 (0.... In PAGE 33: ...ercent. Excluded groups are Antsiranana and and population age 15-49. (the capital province), Fianarantsoa, and Toliara more unequal than the urban areas in the rest of the country. Table6 e confirms that in Mozambique, too, geographic variables are key indicators of local-level inequality, controlling for population characteristics, mean expenditure levels, and urban/rural differences. Compared with Maputo City, the rest of the country has significantly less inequality.... In PAGE 34: ... Table6 e Correlates of mean Log deviation (GE0) in Mozambique Administrative post-level regression (464 administrative posts) Basic regression + expenditure + urban Percent aged 0-5 -0.002 (0.... ..."

### Table 6d Correlates of mean Log deviation (GE0) in urban Madagascar Firaisana-level regression (131 firaisanas) Basic regression + expenditure

2003

"... In PAGE 28: ... We ask whether inequality levels are correlated with location, controlling for both demographic characteristics of the communities (population size and demographic composition), and mean per capita consumption. Table6 a for rural Ecuador finds strong evidence that inequality in the parroquias of the eastern, Oriente, region is significantly higher than the province of Pichincha in the central, mountainous, Sierra, region. Communities located in provinces in the western, coastal, Costa, region tend to be more equal, significantly so in the provinces of Manabi, Los Rios, Guayas, and El Oro.... In PAGE 29: ... Table6 a Correlates of mean Log deviation (GE0) in rural Ecuador Parroquia-level regression (915 parroquias) Basic regression + expenditure Log population 0.0169 (0.... In PAGE 30: ... Possibly there exists greater heterogeneity in household size in those parroquias with large population shares of young children and that this translates into greater inequality of per capita consumption. In urban Ecuador ( Table6 b) the relatively low inequality in the Costa region is again observed. Relative to the zonas in the capital Quito, inequality in all zonas of the Costa region tends to be significantly lower.... In PAGE 31: ... Table6 b Correlates of mean Log deviation (GE0) in urban Ecuador Zona-level regression (660 zonas) Basic regression + expenditure Log population -0.013 (0.... In PAGE 32: ... Table6 c Correlates of mean Log deviation (GE0) in rural Madagascar Firaisana-level regression (1,117 firaisanas) Basic regression + expenditure Log population 0.010 (0.... In PAGE 32: ...55 when all variables are included. In urban Madagascar, the explanatory power is even greater ( Table6 d). Here, unlike rural areas, population size is significantly negatively associated with inequality.... In PAGE 33: ...ercent. Excluded groups are Antsiranana and and population age 15-49. (the capital province), Fianarantsoa, and Toliara more unequal than the urban areas in the rest of the country. Table6 e confirms that in Mozambique, too, geographic variables are key indicators of local-level inequality, controlling for population characteristics, mean expenditure levels, and urban/rural differences. Compared with Maputo City, the rest of the country has significantly less inequality.... In PAGE 34: ... Table6 e Correlates of mean Log deviation (GE0) in Mozambique Administrative post-level regression (464 administrative posts) Basic regression + expenditure + urban Percent aged 0-5 -0.002 (0.... ..."

### Table 6e Correlates of mean Log deviation (GE0) in Mozambique Administrative post-level regression (464 administrative posts) Basic regression + expenditure + urban

2003

"... In PAGE 28: ... We ask whether inequality levels are correlated with location, controlling for both demographic characteristics of the communities (population size and demographic composition), and mean per capita consumption. Table6 a for rural Ecuador finds strong evidence that inequality in the parroquias of the eastern, Oriente, region is significantly higher than the province of Pichincha in the central, mountainous, Sierra, region. Communities located in provinces in the western, coastal, Costa, region tend to be more equal, significantly so in the provinces of Manabi, Los Rios, Guayas, and El Oro.... In PAGE 29: ... Table6 a Correlates of mean Log deviation (GE0) in rural Ecuador Parroquia-level regression (915 parroquias) Basic regression + expenditure Log population 0.0169 (0.... In PAGE 30: ... Possibly there exists greater heterogeneity in household size in those parroquias with large population shares of young children and that this translates into greater inequality of per capita consumption. In urban Ecuador ( Table6 b) the relatively low inequality in the Costa region is again observed. Relative to the zonas in the capital Quito, inequality in all zonas of the Costa region tends to be significantly lower.... In PAGE 31: ... Table6 b Correlates of mean Log deviation (GE0) in urban Ecuador Zona-level regression (660 zonas) Basic regression + expenditure Log population -0.013 (0.... In PAGE 32: ... Table6 c Correlates of mean Log deviation (GE0) in rural Madagascar Firaisana-level regression (1,117 firaisanas) Basic regression + expenditure Log population 0.010 (0.... In PAGE 32: ...55 when all variables are included. In urban Madagascar, the explanatory power is even greater ( Table6 d). Here, unlike rural areas, population size is significantly negatively associated with inequality.... In PAGE 33: ... Table6 d Correlates of mean Log deviation (GE0) in urban Madagascar Firaisana-level regression (131 firaisanas) Basic regression + expenditure Log population -0.014 (0.... In PAGE 33: ...ercent. Excluded groups are Antsiranana and and population age 15-49. (the capital province), Fianarantsoa, and Toliara more unequal than the urban areas in the rest of the country. Table6 e confirms that in Mozambique, too, geographic variables are key indicators of local-level inequality, controlling for population characteristics, mean expenditure levels, and urban/rural differences. Compared with Maputo City, the rest of the country has significantly less inequality.... ..."