Inequality of Opportunities Among Ethnic Groups in the

Preparing to load PDF file. please wait...

0 of 0
Inequality of Opportunities Among Ethnic Groups in the

Transcript Of Inequality of Opportunities Among Ethnic Groups in the

Philippine Institute for Development Studies
Surian sa mga Pag-aaral Pangkaunlaran ng Pilipinas
Inequality of Opportunities Among Ethnic Groups in the Philippines
Celia M. Reyes, Christian D. Mina and Ronina D. Asis
The PIDS Discussion Paper Series constitutes studies that are preliminary and subject to further revisions. They are being circulated in a limited number of copies only for purposes of soliciting comments and suggestions for further refinements. The studies under the Series are unedited and unreviewed.
The views and opinions expressed are those of the author(s) and do not necessarily reflect those of the Institute.
Not for quotation without permission from the author(s) and the Institute.
December 2017
For comments, suggestions or further inquiries please contact: The Research Information Department, Philippine Institute for Development Studies 18th Floor, Three Cyberpod Centris – North Tower, EDSA corner Quezon Avenue, 1100 Quezon City, Philippines Tel Nos: (63-2) 3721291 and 3721292;  E-mail: [email protected]
Or visit our website at

Inequality of opportunities among ethnic groups in the Philippines
Celia M. Reyes, Christian D. Mina and Ronina D. Asis Abstract
This paper contributes to the scant body of literature on inequalities among and within ethnic groups in the Philippines by examining both the vertical and horizontal measures in terms of opportunities in accessing basic services such as education, electricity, safe water, and sanitation. The study also provides a glimpse of the patterns of inequality in Mindanao. The results show that there are significant inequalities in opportunities in accessing basic services within and among ethnic groups in the Philippines. Muslims (particularly the IPs) are the worst-off ethnic groups while the nonindigenous/non-Muslim groups are the better-off groups. Disparities in terms of literacy rate and access to electricity and sanitation between ethnic groups, however, appear to be narrowing between 2000 and 2010.
1 Introduction
Income inequality has continued to persist even in Asian economic giants1 like Singapore and China albeit considerable reduction in absolute poverty. For the past two decades, income inequality in the East Asian region2 has risen by over 20 percent, which largely contributed to persistence of poverty in the region (NEAT, 2015). In the case of the Philippines, income inequality has been following a generally downward trend since 1998. After reaching its peak at 0.5183 in 1997 (during the height of the Asian Financial crisis), the Gini coefficient had consistently been going down from 0.5045 in 2000 to 0.4714 in 2012—the lowest point so far during the covered period of 1991-2012 (Figure 1). This downward trend largely reflects the income distribution in urban areas. On the other hand, income distribution in rural areas has been on the rise since 1991. Periods of rising inequality in rural areas are 1994-1997 and 2009-2012. Arguably, this can be attributed to the bias towards urban and coastal areas but against rural and inland regions due to emergence of new economic opportunities brought by technological change, globalization and market-oriented reforms (Yap, 2013). Decile dispersion ratio has also not significantly reduced for almost three decades. Income of the richest decile has remained around 20 times of the income of the poorest decile (Figure 2). As a result, the poverty situation in the country has not significantly improved and geographical disparity still exists.
 Senior Research Fellow, Supervising Research Specialist and Senior Research Specialist, respectively, at the Philippine Institute for Development Studies. The authors acknowledge the excellent research assistance provided by Ms. Maria Blesila D. Mondez and Mr. Arkin A. Arboneda, Senior Research Specialist and Research Analyst II, respectively. The authors are also grateful for the assistance extended by the National Commission on Indigenous People (NCIP) in coming up with the major ethnic group classification and for sharing some relevant materials and the Philippine Statistics Authority (PSA) for the census data sets.1 Mukhopadhaya et al. (2011) tagged China and India as the Asian demographic and economic giants, while Li and Xu (2016) considered Singapore as one of the four Asian economic giants, together with Hong Kong, Taiwan and South Korea. 1 Mukhopadhaya et al. (2011) tagged China and India as the Asian demographic and economic giants, while Li and Xu (2016) considered Singapore as one of the four Asian economic giants, together with Hong Kong, Taiwan and South Korea. 2 ASEAN+3 countries, composed of Brunei Darussalam, Cambodia, China, Indonesia, Japan, Republic of Korea, Laos, Malaysia, Myanmar, Philippines, Singapore, Thailand, and Viet Nam.

Figure 1. Gini coefficient, Philippines, 1991-2012, by area

0.51 0.49 0.4803
0.47 0.45 0.4736
0.39 0.37 0.3941
0.35 1991

0.4735 0.4602
0.3942 1994









0.4871 0.4513 0.4288

0.4837 0.4496 0.4296

Source: Reyes et al. (2012)

All Areas


Figure 2. Decile dispersion ratio, Philippines, 1985-2012


0.4743 0.4462 0.4278

0.4714 0.4476 0.4471






0 1985 1981 1991 1994 1997 2000 2003 2006 2009 2012
Source of basic data: 2000-2012 Family Income and Expenditure Surveys, Philippine Statistics Authority
Inequality is not all about inequality of outcome (which is commonly measured by income or consumption), or inequality caused by differences in “effort,” which is referred to as “the choice variable for which a person should be held responsible” (Romer, 1998; as cited Kanbur, 2014, p. 6). There is another component of inequality, which is termed as the inequality of opportunity and is caused by differences in [exogenous or uncontrollable] “circumstances,” or “attributes of a person’s environment for which he should not be held responsible” (Romer, 1998; as cited Kanbur, 2014, p. 6). Inequality of opportunity is considered unacceptable under the egalitarianism principle, should be reduced and should inform the public policy design (Kanbur, 2014; Son, 2013). Therefore, other than income inequality, it is also interesting to examine inequality using non-income-based indicators such as access to education and other basic services, especially those among ethnic groups.
There has been very little work done examining inequalities among and within ethnic groups in the Philippines. This is primarily because data on characteristics of the 180 ethnic groups in the country is very scant. The main source of information is the census of population and housing conducted by the Philippine Statistics Authority (PSA; formerly the National Statistics Office) conducted every 10 years.

This study aims to show inequalities among different ethnic groups in the Philippines. This is part of a research initiative of UNU-WIDER to address the measurement of horizontal inequalities in developing countries. The paper examines inequality in opportunities in accessing basic services. In particular, this paper looks at access to education as measured by years of schooling and literacy rate, and access to basic amenities as measured by access to safe water, sanitation and electricity.
The Philippines is composed of three major island groups – Luzon in the north, Visayas in the middle, and Mindanao in the south. A quarter of the country’s total population resides in Mindanao. Parts of Mindanao have been plagued by conflict and this has been linked to religious conflicts (Muslims vs. Christians) as well as the clash of interests in land and other natural resources (affecting indigenous populations). Thus, this paper also examines the patterns of inequality in Mindanao.
2 Review of literature
The empirical literature on inequality focused more on income-based measures and the concept of vertical (or within-group) inequality. Some of the notable studies that utilized the Philippine data include Kanbur and Zhuang (2013) and Balisacan and Fuwa (2004). Kanbur and Zhuang (2013) noted that the national-level income inequality has inched up due to urbanization and rising rural inequality. Balisacan and Fuwa (2004), on the other hand, argued that the national-level income inequality is largely due to income differences within the region. Regional differences, or the so-called spatial inequality, account only for a small component of inequality. The study also mentioned that there had been income convergence among provinces probably due to human capital stock and land distribution, among others.
Horizontal inequality among different ethnic groups in terms of non-income-based indicators has not yet been explored much in the literature. A few studies that tackled topics related to this include those of Stewart et al. (2010), McDoom and Gisselquist (2015), Lindquist (2011), and Selway (2011), among others.
Some studies focused on the measurement and monitoring of horizontal inequality. Stewart et al. (2010) proposed a methodology of measuring and monitoring the horizontal inequality and demonstrated it using longitudinal income data from South Africa and census data from Indonesia. The study concluded that group-weighted coefficient of variation, group-weighted Gini and groupweighted Theil’s index are all suitable measures of horizontal inequality. An earlier study by Stewart (2009) defined horizontal inequality and illustrated its presence using the 1995 inter-censal survey data from Indonesia. The results suggest that political as well as cultural status inequality (which leads to violent unrest) exists in countries where Muslims form a minority. In countries where Muslims form a majority, on the other hand, economic inequalities are compensated for by political power and cultural status. It also provided evidence on the international links across Muslim groups. Selway (2011), on the other hand, introduced the concepts of crosscuttingness and cross-fractionalization. In relation to this, Abanes et al. (2014) examined the relationship between ethno-religious categorization, identification and social distance by testing the mediation of out-group trust using the Philippine data. The study randomly surveyed university students in Metro Manila and Autonomous Region in Muslim Mindanao (ARMM). The study revealed that there are significant differences by ethno-religious categorization on social distance. In addition, it has been found that people who strongly identify with their religion tend to maintain social distance with religious out-groups, and this can be explained by out-group trust. McDoom and Gisselquist (2015) estimated various measures of ethno-religious divisions (e.g., horizontal inequality, fractionalization, crosscuttingness) for Mindanao using the 2000 and 2010 individual-level census data for the Philippines. The analyses suggest that horizontal

inequalities between ethnic groups can explain the nexus between ethnic divisions and ethnic civil war as well as that between ethnic divisions and less provision of public goods.
Some studies specifically estimated inequality of opportunity. Son (2013) presented a measure of inequality of opportunity—the Human Opportunity Index (HOI)—using household-level survey data from seven developing countries, including the Philippines. Findings of the study include: (1) inequality in terms of primary school attendance is higher than that of secondary school attendance; (2) main factors affecting inequality of opportunity for education are per capita household expenditure, location and education of household head; (3) inequalities in terms of access to basic infrastructure services like safe water, electricity and sanitation are lower; and, (4) main factor affecting inequality of opportunity in terms of access to safe water and sanitation is per capita household expenditure. Marrero and Rodriguez (2012) measured inequality of opportunity and compared the estimates across European countries. The study also identified the set of characteristics with causal effect on inequality of opportunity. Using the 2005 cross-sectional data for 26 European countries, the study revealed that countries with low inequalities are Nordic, continental and some Eastern countries while countries with high inequalities are the Mediterranean, Atlantic and other Eastern countries. It has also been found that total social protection expenditure, dropping out from school, reaching secondary level education, as well as development and labor market variables negatively correlate with inequality of opportunity. Singh (2012) also estimated inequality of opportunity in earnings and consumption expenditure for different aged-based cohorts in India using both parametric and nonparametric approaches. Nonparametric approach revealed that inequality of opportunity in earnings is lower in rural areas than in urban areas, and significant factors affecting inequality includes absence of high paying jobs in rural areas and limited choices regarding decisions about their children due to infrastructural constraints in rural areas. Results from the parametric approach include the following: (1) father's education and occupational status have positive effect on earnings and consumption expenditure; (2) father's education has higher maximum opportunity share in earnings inequality in urban areas than in rural areas; and, (3) opportunity shares of circumstances are relatively larger in rural areas. Moreover, Ferreira and Gignoux (2011), on the other hand, introduced the absolute and relative versions of the lower-bound index of inequality of opportunity. The study noted that inequality of opportunity ratios are higher for consumption than for income while inequality of opportunity levels are generally lower for consumption and for income. Opportunity deprivation is also found to be strongly correlated with ethnicity, region and family background.
There are also studies that looked into possible relationship between conflicts and inequality. Using pooled cross-section data on European countries, Lindquist (2011) found that horizontal inequality (in terms of access to education) can significantly predict the occurrence of ethnic and civil conflict. Caprioli (2005) examined the impact of gender inequality on the probability of intrastate conflict using PRIO/Uppsala data set of internal conflict. The study found that higher levels of gender inequality within a state has higher probability of experiencing internal conflict. In addition, presence and number of at-risk minorities, transitional polities and prior conflict increase the probability of internal conflict. Vinck (2011) is one of the local studies that looked into the violent conflicts in Central Mindanao. Based on a series of interviews conducted in selected areas in mainland Mindanao, the study found that violent conflict in Central Mindanao has caused mass displacement between 2000 and 2010. Another local study is Edillon (2005), which examined the determinants of incidence of armed conflicts in the Philippines using the time-series data on armed conflicts for the period 19722004. Some of the key findings of the study include the following: (1) the most significant determinant of incidence of conflict is government’s policy on peace and income redistribution; (2) deprivation in access to water is a considered as a major cause of conflict; (3) minoritization and average permanent income are positively correlated with incidence of conflict.

3 Methodology

3.1 Data

The main sources of data are the Censuses of Population and Housing (CPH) conducted by the PSA in 2000 and 2010. The CPH has a long form that collects a few demographic and social information on the characteristics of the population and this is administered to at least one tenth of the population. The data for the 10 and 20 percent samples of the 2000 and 2010 CPH, respectively, were used in this study. Around 7 million individuals in 2000 and close to 20 million individuals in 2010 were processed to generate the measures of inequality across ethnic groups. It would have been ideal to examine economic disparities among ethnic groups but the CPH does not collect such data. Thus, this study can only examine outcome indicators that are available in the CPH.

Data on other indicators are sourced from administrative records from different government agencies.

3.2 Variables

Outcome variables

The outcome variables considered in this paper are average years of schooling (among those aged 25 and over), literacy status (among those aged 10 and over), access to safe drinking water, access to sanitary toilet facility, and access to electricity. These non-income indicators are believed to be strongly correlated with income and welfare status. The definition of these variables are presented in Table 1.

Table 1. Definition of outcome variables

Variable Schooling Literacy status
Access to safe water

Definition average years of schooling of an individual aged between 25 and over
1 if an individual aged between 10 and over is literate (or can both read and write a simple message); 0 if illiterate
1 if an individual belongs to a household having an access to a safe drinking water (or if main source of drinking water supply is either community water system, tubed/piped well or bottled water); 0 otherwise

Access to sanitary toilet facility Access to electricity

1 if an individual belongs to a household having an access to a sanitary toilet facility (or if type of toilet facility is either water-sealed sewer septic tank/other depository or closed pit); 0 otherwise
1 if an individual belongs to a household having an access to electricity; 0 otherwise

Grouping variables
The grouping variable is one of the main considerations in estimating inequality, particularly inequality of opportunities.
There are three grouping variables used in this study. One of these, and is the most important one, is ethnicity. Ethnicity is a primary sense of belonging to an ethnolinguistic group, which is consanguineal in nature in the sense that the ties are reckoned by blood and traced through family tree (PSA, 2016b). Ethnic grouping in the Philippines denotes genealogical, paternal as well as maternal lineage to any of the country’s group of native population3 (PSA, 2016a). The Philippines has a total of 182

3 Maternal lineage has been included for the purpose of census (PSA, 2016a).

ethnolinguistic groups; around 110 of which are considered as indigenous people (IP) groups. As defined in the Indigenous Peoples Rights Act (IPRA) of 1997, IPs are referred to as follows:

“a group of people or homogeneous societies identified by self-ascription and ascription by others, who have continuously lived as organized community on communally bounded and defined territory, and who have, under claims of ownership since time immemorial, occupied, possessed and utilized such territories, sharing common bonds of language, customs, traditions and other distinctive cultural traits, or who have, through resistance to political, social and cultural inroads of colonization, non-indigenous religions and cultures, become historically differentiated from the majority of Filipinos[; or] peoples who are regarded as indigenous on account of their descent from the populations which inhabited the country, at the time of conquest or colonization, or at the time of inroads of nonindigenous religions and cultures, or the establishment of present state boundaries, who retain some or all of their own social, economic, cultural and political institutions, but who may have been displaced from their traditional domains or who may have resettled outside their ancestral domains.” (IPRA, Chapter II, Section 3h; as cited in ADB, 2002)

Since there are more than a hundred (i.e., 147 and 182 in 2000 and 2010, respectively) ethnolinguistic groups in the Philippines that are reported in the CPH, the authors decided to create major groups out of these many smaller ethnic groups. Based on the classification used by the NCIP, this study came up with three major ethnic groups, namely: (1) Muslim ethnic groups; (2) Indigenous nonMuslim ethnic groups, or non-Muslim IPs; and, (3) Non-indigenous/non-Muslim ethnic groups, or non-Muslim/non-IPs. The first group is composed of ethnic groups that are Muslims—also known as Moros in other studies; regardless of whether they are IPs or not. It has two sub-groups—the indigenous Muslim ethnic groups and the non-indigenous Muslim ethnic groups. According to the NCIP, the indigenous Muslim ethnic groups are those that embrace the Islamic faith and, at the same time, continue to practice their own culture and tradition as IPs. The non-indigenous Muslim ethnic groups are not classified as IPs by the Office of the Muslim Affairs (OMA) but profess the Islamic faith. In 2010, this group comprised the following small ethnic groups/tribes:

Table 2. List of Muslim ethnic groups

Indigenous Muslim ethnic groups
1. Badjao 2. Iranon/Iranun/Iraynon 3. Jama Mapun 4. Kalagan 5. Kalibugan/Kolibugan 6. Sama Badjao 7. Sama Bangingi 8. Sama Laut 9. Sama/Samal

Non-indigenous Muslim ethnic groups
1. Maguindanao 2. Maranao 3. Palawani 4. Sangil 5. Tausug 6. Yakan

Source: National Commission on Indigenous Peoples (2010)

The second group are non-Muslim ethnic groups that are officially classified by the NCIP as IPs. In 2010, this major group is composed of 142 ethnic groups nationwide. Refer to Appendix A for the complete list.

The remaining 19 ethnic groups—labeled as “non-indigenous/non-Muslim ethnic groups” by the authors comprised the third major ethnic group (Table 3). These are those that are neither Muslim ethnic groups nor IPs.

Table 3. List of non-indigenous/non-Muslim ethnic groups


Ethnic group

1 Bikol/Bicol

2 Bisaya/Binisaya

3 Boholano

4 Capizeño


5 Caviteño 6 Caviteño-Chavacano 7 Cebuano 8 Chinese 9 Cotabateño 10 Cotabateño -Chavacano 11 Davao-Chavacano 12 Davaweño 13 Hiligaynon/Ilonggo 14 Ilocano 15 Kapampangan 16 Masbateño/Masbatenon 17 Pangasinan/Panggalato 18 Tagalog 19 Waray
Source: National Commission on Indigenous Peoples (2010)
Moreover, the second and third grouping variables used in this study are religion and language/dialect generally spoken at home. These variables are important in examining the homogeneity of different ethnic groups in terms of religion and dialect.
From around 82 and 97 religious groups in 2000 and 2010, respectively, five major groups were generated in this study based on the categories used in Pew Research Center (2015). These are the following: (1) Roman Catholic; (2) Muslim; (3) Other Christians (i.e., Protestant, Church of Jesus Christ of Latter-day Saints or Mormon, Jehovah’s Witness, others); (4) Tribal/indigenous religion; and, (5) Other non-Christians (i.e., Jewish, Buddhist, Hindu, others). Christians are defined here as those who believe in the Holy Trinity and that Jesus Christ is God. Mormons and Jehovah’s Witnesses, both originated in the United States, are categorized under Other Christians albeit their departures from traditional Christian beliefs (as they have their own interpretations of the Bible and own view of the Holy Trinity) (Pew Research Center, 2011). Iglesia ni Cristo, on the other hand, is considered as a non-Christian religious group since its set of beliefs is categorized under the Unitarian (Universalist) faith (Pew Research Center, 2015).
Meanwhile, there are as many languages/dialects generally spoken at home as ethnic groups in the country. Five major categories were used in this study, and these are the following: (1) Tagalog; (2) Other major languages/dialects in Luzon (i.e., Ilocano, Bikol/Bicol, Kapampangan, and Pangasinan/Panggalato); (3) Major languages/dialects in Visayas (i.e., Hiligaynon/Ilonggo, Cebuano, Bisaya/Binisaya, Waray, Karay-a, Boholano); (4) Major languages/dialects in Mindanao (i.e., Maguindanao, Maranao, Tausug, Surigaonon, Zambageño-Chavacano, Sama/Samal); and, (5) Other languages/dialects. The dialects belonging to the second, third and fourth major groups are selected based on their distribution. For instance, Ilocano, Bikol/Bicol, Kapampangan, and Pangasinan/Panggalato are the four most commonly used spoken dialects in Luzon, next to Tagalog. At least 1.3 million Filipinos who are living in Luzon speak these dialects. On the other hand, Hiligaynon/Ilonggo, Cebuano, Bisaya/Binisaya, Waray, Karay-a, and Boholano are the largest dialect groups in Visayas, with at least 870,000 speakers. The first four dialects have more than 2 million speakers in Visayas. Other than Bisaya/Binisaya, Cebuano and Hiligaynon/Ilonggo that are also being spoken in Mindanao, the authors identified six major dialects being used in Mindanao, with at least 300,000 speakers. These are Maguindanao, Maranao, Tausug—with at least 1 million speakers each— , Surigaonon, Zambageño-Chavacano, and Sama/Samal.
3.3 Inequality measures
Different measures of inequality are estimated to determine whether there is an unequal access to basic services across different groups (“between-group”) and across members of each group (“within-

group”). The most common of these measures are Gini coefficient, Theil’s index and coefficient of variation4. This study also presents measures of crosscuttingness and cross-fractionalization proposed by Selway (2011) as well as the HOI developed by the World Bank.
Gini coefficient
The Gini coefficient5 is the most commonly used inequality measure. Its values range from 0 to 1, indicating perfect equality and perfect inequality, respectively. This measure can be computed using the following equation:

  G  1 1  


 





N  yN 2 


where persons are ranked in ascending order of yi . This measure cannot usually be written as the sum of a term summarizing within-group inequality and a term summarizing between-group inequality. Consider a population of persons (or households), i = 1, 2, …, n, with outcome variable yi and wi .

f  wi i N, where

N   wi .
[In what follows all sums are over all values of whatever is subscripted.] Arithmetic mean income is y . Suppose there is an exhaustive partition of the population into mutually-exclusive subgroups k = 1, 2, …, K.
Theil’s index
The Theil’s index belongs to the Generalized Entropy class of inequality indices, which is given by the following formula:

 GE(1) 


 


  log


 

 y   y .

This index, which ranges from 0 to log n, can be additively decomposed as follows:

4 These can be estimated using the following commands in the Stata software: ineqdeco, ginidesc, ainequal, egen_inequal, and iop. The first two commands provide between- and within-group components. The last one is commonly used when the variable of interest is dichotomous or binary.
5 Formulas for Gini coefficient and Theil’s index were mainly sourced from Stata’s help desk on ineqdeco.

GE(1)  GEW (1)  GEB(1) ,

where: GEW (1) is the ‘within-group’ inequality while GEB(1) is the ‘between-group’ inequality. Furthermore,

GEW (1)


(1 k




a k





v  Nk kN

is the number of persons in subgroup k divided by the total number of persons (subgroup population share), and sk is the share of total income held by k’s members (subgroup income share). (Strictly speaking, vk is the sum of the weights in subgroup k divided by the sum of the weights for the full estimation sample.)

GEk(1) , which is the inequality for subgroup k, is calculated as if the subgroup were a separate population, and GEB(1) is derived assuming every person (or household) within a given subgroup k received k’s mean income, yk .

Group-weighted coefficient of variation

The group-weighted coefficient of variation (GCOV) is given by the following formula:


   1  R
GCOV  



y 2

2 

y r r r


 1 nr

yr 

yir is the group r’s mean value;

nr i

R is the number of groups;

pr is group r’s population share;

yir is the quantity of the variable of interest (e.g., years of education) of the ith member of group r

The coefficient of variation is a common measure of regional disparities. GCOV is weighted by the population size of each group, so that changes in the position of small groups get less weight than those of larger groups (Mancini, 2005).