Marginality as a root cause of urban poverty: a case study of Punjab.
Zahra, Kanwal ; Zafar, Tasneem
Marginality is a complicated patchwork of various aspects and is
deep rooted in institutional factors and resource constraints. It refers
to a position where in an individual or group is at margin of
socio-economic and political system due to some causal complexes and
interplays that exclude some segments of the population from the growth
prospects. A consensus has emerged in recent literature on poverty and
marginality, where marginality has been viewed as a root cause of
poverty and its perseverance. This paper is very much in line with this
theoretical conception and marginality is identified here as patterns of
causal factors in societal, economic and spatial dimensions that shape
the root causes of poverty and capture marginalised poor in a vicious
circle from which they want to escape. The urban context is the
particular focus of this paper. On the basis of the analytical framework
developed, study examines the relationship of marginality and poverty in
a systematic manner and investigates the multidimensional poverty among
marginal classes living in the urban areas of the Punjab province of
Pakistan. To this end, the study makes the use of data from two waves of
Multiple Indicator Cluster Survey (MICS) for the year 2007-08 and
2010-2011 and draws results from Poison regressions which reveal a
strong impact of marginality and exclusion on being multidimensional
poor.
JEL Classification: 131,132
Keywords: Marginality, Multidimensional Poverty, Exclusion
I. INTRODUCTION
Historically poverty as a concept was considered to be a key factor
to design social policy. The social development normally is concerned
with socio-economic empowerment of the poor of the concerned society. It
is always been a key issue for developing as well as developed
countries, however the nature and treatment of issue varies. The
treatment of poverty is different from society to society. In advanced
countries, an individual who is unable to actively participate in
society or has weak social network, environment, health and education
etc. is considered to be poor. Financial empowerment is also considered
to be important in these countries but it takes into account with other
dimensions of poverty [Lyberak and Tinios (2005)]. However, in
developing countries, policy focus is still on uni-dimensional
definition of poverty where a single dimension either consumption or
income is a strong factor that affects the standard of living of an
individual [Wagle (2005)]. Though the multidimensional poverty concept
is also getting attractiveness in these countries with a perception that
an individual' status in one dimension cannot represent his status
in another dimension but still there remains dearth of policy-making.
Another important transformation in the literature on poverty is seen in
terms of identification of nexus of marginality, social exclusion and
poverty [Ruth, et al. (2007); Zoran, et al. (2006); Whelan and Bartrand
(2005)].
In developed countries, marginality is being treated as phenomena
related with poverty and social exclusion. Separate surveys were
conducted to see the root cause of the problem, in Canada, Canadian
Institute of health sciences introduced marginality index as a policy
measure. In developing countries, unfortunately very limited literature
is available in the area of marginality and social exclusion. However in
India, due to caste inequalities, this issue is getting great attention
of the researchers [World Bank (2011); Thorat and Nidhi (2010); Thorat,
et al. (2009); Mitra (2004)]. Marginality is broadly defined as a state
situated at the margin, this could lead toward social exclusion and
hence poverty. On the other hand the term "social exclusion"
is a vibrant, multidimensional process driven by unequal power
relationships. This exclusion can affect individual, household, group,
community and countries across four dimensions i.e. economic, political,
social and cultural and make certain objects more vulnerable which leads
them to high incidence of poverty [Jennie, et al. (2008)]. In this
respect, the study of poverty dynamics could benefit from engaging with,
and incorporating, models or detailed conceptions of multidimensional
social exclusion among the marginalised population.
The state of poverty among this marginalised class may be different
from the rest of the population. This marginalised class may face
exclusion in socio-economic and cultural participation in the society
which deprives them from education, health, networking etc. However, the
determinants of poverty may be the same but the effect of these
determinants could be different. A number of literature is also
available which theoretically and conceptually establishes dimensions
and characteristics of marginalisation, addressing processes of
restricted participation of this marginalised class in social, economic,
political or cultural life of common society. Poverty and social
exclusion as two descriptions of severe social inequality have often
been mixed up, and hypothesis about their interrelation and
characteristics have hardly been tested empirically. Is poverty the
first stage on the way out of society, or are there considerable
differences between the risks of becoming poor or socially excluded? To
this end, this study proposes the conceptualisation and
operationalisation of social exclusion tendencies and incidence of
poverty in the identified marginalised class. The empirical analysis
employs Poisson regression analysis to see the determinants of
multidimensional poverty among the identified marginalised group.
Organisation of paper is as follows: Section I is introduction.
Section II presents literature review. Theoretical framework is
discussed in Section 111. Section IV throws light on methodology.
Empirical model, estimation techniques and explanation of variables is
there in Section V of the paper which is followed by discussion of
results in Section VI. Section VII concludes the paper.
II. LITERATURE REVIEW
The rapid rural-urban migration, structural changes in developing
economies and globalisation is enhancing the issue of urban poverty and
is creating serious problems in the management of urban areas of
developing countries, Pakistan being one of them. One of the evident
causes of poverty and inequality in the urban areas is marginality
[Susan, et al. (2001); WDR (2001); Oxfam GB Urban Programme (2009)].
Poverty is not a unidimensional concept and is not the name of material
deprivation but is a set or an outcome of interlocking factors such as
physical weakness, socio-economic isolation, vulnerability and
powerlessness [Philip (2008)]. This poverty type may be same or
different within certain socially excluded groups and is strongly
dependent upon the clan network of households existing in a marginalised
group. Unfortunately these factors leading to poverty have received less
attention of the researchers in Pakistan, mostly research on the issue
of poverty in Pakistan explores levels, trends and dynamics but not much
attention is given to the issue of vulnerability to multidimensional
poverty of the marginalised livelihood of this country.
A person is normally considered deprived due to deficit consumption
but there are other factors that contribute significantly to make one
feel deprived including the shortfall of living needs. The living
standards are highly affected by insecurity and powerlessness of future
shortfalls. Calvo (2008) therefore considered this vulnerability to
multidimensional poverty as a form of hardship that is defined in both
conceptual and empirical way. He extended his own index that he
developed in 2005 and used bidimensional measures of consumption and
leisure. His findings suggest that these two dimensions are negatively
correlated in both rural and urban cases. This vulnerability is
different from poverty much attention is needed to differentiate between
vulnerability and poverty.
Vulnerability is related with poverty but it is not necessary that
all poor are vulnerable or all vulnerable are poor. Angemi (2011)
supported this view in his study with the help of household level
analysis within poverty framework. He pointed out that the
characteristic of vulnerability is consistent with the characteristics
of poor so by this he found that poverty and vulnerability both are
related with each other. However, an important point of his analysis was
that all poor are not vulnerable while some proportions of non-poor are
vulnerable. In the same lines Susan and Takashi (2002) employed two
period panel data set of the North-West Frontier Province, Pakistan and
proved that the sample household was subject to a high risk of income
poverty. Results also revealed the households are more vulnerable to
consumption poverty and are affected by the shock of outside employment
as compared to self-employed households. An important outcome from this
analysis concludes that the age, having less land and irregular sources
of income strongly affect the extent of vulnerability among households.
Diego (2011) is of the view that the dynamics of risk and uncertainties
are helpful to understand the nature of poverty. By applying the pooled
GLS method on the national data sets of Uganda, he discovered that along
with a sharp reduction in poverty, the vulnerability to poverty in
Uganda has also declined, however, the issue of marginalisation existed
due to geographical segregation. The results revealed that the central
region experienced reduction in incidence of vulnerability while the
rural areas, where 90 percent of population is living under extreme
poverty conditions, the incidence of vulnerability has increased.
Supporting the findings of Diego (2011), a worldly accepted truth is
that this high incidence of vulnerability to poverty is mostly dominant
in socially excluded and marginalised group.
Early research also support the idea of this social exclusion, In
industrialised countries, the evolution of one parent family defines a
new pattern of poverty and marginalisation. This marginalisation exists
not only in labour market of these countries but also exists in the
provision of public housing [Hilary (1989)]. On the other hand, David,
et al. (2000) tried to develop a baseline for understanding the nature
of poverty and social exclusion. They used poverty in terms of
deprivation from goods, services and social activities. They are of the
view that this way of measuring deprivation satisfied both absolute and
relative poverty terms. The analysis shows the there is an increase in
the multiple deprivation and poverty in Britain during the survey
period. By identifying these issues in family-cycle approach, Dewilde
(2003) tried to develop a framework of analysis of poverty and social
exclusion. As per his views, a life course perspective reconceptualises
the traditional approaches and combines their best element into the
analysis of social exclusion hence poverty. He used three sociological
perspectives on the life course i.e. the traditional North-American life
course perspective by Elder (1974), the Continental institutional
approach and "political economy of the life course". With the
help of these three approaches, he proposed a new framework to analyse
poverty and social exclusion relationship over the life course, both
theoretically and empirically.
These circumstances of poverty are strongly related with level of
social exclusion and parental social class. The factors that provide the
poverty prospects at childhood age due to parental social class are
strongly associated with current lacking of basic infrastructure [Aya
(2009)]. This was also proved by Christopher, et al. (2013) with the
help of a comparative analysis between four important factors i.e.
social exclusion, parental status, childhood economic status and state
of current poverty. With the help of EU-SILC module, they figured out
how the welfare regimes mediate the impact of parental social class and
childhood economic circumstances on poverty. Findings showed that by
applying social class plan, intergenerational factors have least impact
on income poverty. The other objective of the analysis was to get
knowledge about the impact of parents' class and childhood economic
circumstances on income poverty and it was discovered that the impact of
parent social class on income poverty is weak for social democratic
countries and strong for liberal countries, however social class has
high impact on vulnerability. In case of income poverty the impact of
vulnerability is high in relation to both parent's social class and
childhood economic circumstances. Economic vulnerability has also high
impact on welfare regimes which experienced difficult economic
circumstances in childhood.
Franz, et al. (2011) has provided conceptual and analytical
framework in order to explore the root cause of poverty. They were of
the view that there is need to highlight poverty with respect to
marginality. They found marginality to be the root cause of extreme
poverty. According to them, marginality is an involuntary position and
is a condition of an individual or group that is at the brink of social,
economic and ecological systems. Such marginality prevents affected
communities to utilise resources, assets and service and all other
factors, that become the cause of poverty. They define poverty as a
matter of absolute deficiencies as perceived by the poor. They look
poverty as a relative, subjective, dynamic and systematic mechanism and
concluded marginality as a pattern of causal complexes in a societal and
spatial dimension.
Nayar (2007) is of the view that poverty and social exclusion that
are significant socio-economic variables and are generally ignored while
estimating ill-health effects. Social exclusion mainly refers to the
inability of a society to realise its full potential while keeping all
groups and individuals within reach. The relationship between caste and
health indicator shows that poverty is a complicated issue that requires
to be addressed with a multi-dimensional facet.
Literature no doubt covers issues of poverty, marginality and
social exclusion on very broad way. Valuable input was given by
different authors to explore issues of poverty among marginalised class.
But there is a lack of literature available in identification of
marginal and socially excluded population from poverty and social status
dataset. Researchers made effort by conducting survey of marginalised
group but that was not at province or national level, however the
importance of defining socially excluded class at national level is
vibrant. Therefore this study identifies marginal population from
existing dataset and analyses extent of their multidimensional poverty
in Punjab.
III. THEORETICAL FRAMEWORK
Poverty is a long term debate and developing countries are
targeting to be free of poverty by 2015, the millennium development
goals directly and indirectly target poverty eradication and aims for a
good standard of living for the livelihood of the society. To eradicate
extreme poverty and to make people out of extreme hunger requires a good
educational infrastructure; reduced child mortality, improved maternal
health and gender equality and enhanced women empowerment [United Nation
(2007)]. Progress towards reducing poverty is slower which addresses
policy gaps in achieving the target. Policies overlook the depth in the
issues of poverty and take poverty at general level, but the population
who is actually excluded from rest is ignored, that population is living
below poverty line and marginalised in participating socio-economic
activities with rest of the population of the region. Unfortunately
pro-poor growth ignores this important aspect of poverty. The facts
shows progress is slower in developing world where globalisation is seen
in form of higher rural-urban migration but on other side, the economic
and social side is still deprived and fails to meet the challenges of
this higher rate of rural-urban migration. This causes an increase in
the burden of city management and also an increase in the size of the
excluded area within the city or periphery of the city. Such population
is marginalised while living in the slums and katchiabadies of urban
area and face a lack of opportunities to acquire skills and access to
labour market. This marginalised population then becomes socially and
ethnically excluded from the rest of the society and has less access to
educational, health and other urban services.
[FIGURE 1 OMITTED]
This marginalisation defines boundaries between groups living in a
society, some groups are economically excluded and to some extent social
inclusion prevail in such group, but on the other hand some are
demographically and economically excluded, in a society of developed as
well as developing countries, therefore marginalisation can be
considered as a process in which a community or individual lives at
margin and gradually become economically, culturally, socially and
politically excluded from rest of population [Zahra and Tasneem (2014)].
There are some deprived groups who are excluded in all dimensions of
exclusion and spent deprived and vulnerable life even being part of that
society. Thus marginality leads to social exclusion in long run and this
social exclusion is blamed to be primarily responsible for social
conflict due to its inability to transform itself since it is strongly
connected to the systems of oppression and domination.
Usually poverty links with material lacks, it has theoretical as
well as strategic importance, but the increasing understanding is that
poverty is not just a name of material lacks, but also associated with
restricted access to resources that can make an individual or household
well off. UN has defined poverty through the "capability
approach" and "the human rights approach". These
inter-related themes provide an enriched understanding of poverty and we
can define poverty as:
"A human condition characterised by the sustained or chronic
deprivation of the resources, capabilities, choices, security and power
necessary for the enjoyment of an adequate standard of living and other
civil, cultural, economic, social and political rights". (1)
In continuation of defining poverty, Oxfam (2009) extends this
definition into four dimensions, these dimension includes social
exclusion, relative and income poverty as well as relative poverty
status. Poverty can be defined as:
"Poverty can't be comprehensively defined by a single
approach; it needs to cover the aspects of not having enough to either
live on or to build from and being excluded either from wealth or from
the power to change for betterment, these sums up to four areas."
(2)
Thus poverty cannot be restricted to income and expenditure but it
is the name of deprivation of the resources that makes an individual
better off in his social, economic, cultural and political life. Oxfam
(2009) also explains social exclusion as the fourth dimension of
poverty. Exclusion causes poverty, this relation may be causal and may
make people vulnerable and then poor, this marginality emerges due to
certain groups' representation from ethnic minority, deprived
class, deprived gender and due to lack of participation in social life,
restricted access toward living facilities etc. that causes
vulnerability amongst these groups and in turn poverty in the long run.
This marginality has two way relationships with poverty, it
enforces people to be poor or poverty enforces people to be
marginalised. Individual or settlements being excluded from the
dimension of development and progress move towards extreme poverty. The
people that are affected by poverty and exclusion are considered to be
the marginalised poor [ZEF (2011)].
Poverty is ex post phenomenon of social exclusion, it is caused by
marginality or vice versa. A vulnerable household can be in and out of
poverty over time depending on the future income prospects, expenditure
stream, and accessibility to social services. A marginalised household
can be poor or non-poor. A marginalised household considers being poor
or more sensitive to shocks if that household has:
(i) Low level of human capital, knowledge and access to skill
improvement.
(ii) Suffers from physical or psychological disabilities and poor
access to health facilities.
(iii) Poor infrastructure and have less capacity to improve it.
(iv) Few productive and financial assets and has limited access to
credit market.
(v) Poor social networking and excluded from normal lives of
society.
(vi) Poor access to job market opportunity.
Therefore:
A marginalised household is considered to be poor if it has a
limited access to the living needs, has limited or restricted access to
social, economic and political life of its society due to residential,
societal, spatial, environmental deprivations etc. and has poor capacity
to ensure good standard of living for its members.
IV. METHODOLOGY
The geographical focus of this paper is the Punjab province of
Pakistan, which is an economic hub of the country. The dynamic nature of
agriculture and industrial production along with having major population
share of the country makes it more important than other areas or
provinces. However, Punjab has witnessed major urbanisation in the past
few decades and has achieved improved growth rate but that has not
proved to be beneficial for the entire population and certain segments
of urban areas remain in extreme poverty.
Numbers of studies are available that cover issues of poverty in
Punjab as well as in Pakistan but advanced level analysis on poverty is
rare in literature. Primary data from the combined round of PIHS was
used by Siddiqui (2007) whereas Siddiqui (2009) used PSLM 2004-05
survey. Sikander (2009) used the data from Multiple Indicator Cluster
Survey (MICS)-2003-04 to analyse the determinants of poverty in Punjab.
Malik (1996) used self-collected data on a rural locality called
"Wanda" (District Bhakkar, Punjab). His results were based on
a sample size of 100 and however were not nationally representative for
inference about the determinants of poverty. The analysis of
marginalisation and poverty in this study is based on two waves of data
from Multiple Indicator Cluster Survey (MICS) conducted in 2007-08 and
2011-2012. In 2007-08, 91,280 households participated in the data
collection process out of which 59456 were rural and 31824 were urban.
Of this sample 594,851 individual from urban and rural areas were
covered with a wide range of socio-economic issues on living condition,
economic situation, health and education, housing etc. In data set
collected in 2011-12, 3102048 household were covered, in which 3488 was
urban and 3788 are rural, and this data set also covered more than 90
indicators from different socio-economic perspective. The unit of
observation for the analysis of this study is the individual resides in
the urban areas of Punjab, Pakistan.
This paper is primarily concerned with the measurement of poverty
among marginalised people of Punjab with immediate focus on whether
these people are living in extreme poverty or out of poverty. The
definition behind is that poverty is a relative concern that can be
explained with economic and social wellbeing, capability and social
inclusion. Whereas marginalised and socially excluded concept is another
important dimension of the study which has been extracted from Zahra and
Tasneem (2014). The flow of empirical analysis is based upon
marginalised population as this study is concerned with the measurement
of poverty among marginalised people. Marginalised and socially excluded
population in urban areas of all cities of Punjab are extracted with the
help of an index developed in Zahra and Tasneem (2014)along with the
dimensions and indicators of marginality given below in Tables 1 and 2.
V. EMPIRICAL MODELING, ESTIMATION PROCEDURE AND EXPLANATION OF
VARIABLES
Literature provides different techniques e.g., Ramya, et al. (2014)
and Labar and Bresson (2011) estimated multidimensional poverty index
based on Alkair Foster Measure, whereas Mahlberg and Obersteiner (2001),
Sikander and Mudassar (2008) and Merz and Rathjen (2011) used logit
regression to see multidimensional poverty. Wagle (2005) contributed in
literature by introducing index based estimation of multidimensional
poverty and used structural equation modeling. Literature support a wide
range of methodologies which used structural as well as simultaneous
equation modeling. Attention has now been diverting to analyse the
impact of different deprivation on extent of multidimensional poverty.
The extent of multidimensional poverty can be seen with the help of
number of areas in which a specific household or individual is deprived
[Alkair and Foster (2011); Jhon, et al. (2013)].
Dimensions in which household or individual are deprived is
measured as count data (number of dimension in which each individual is
deprived) and are assessed with Poisson regression, a useful technique
for count data modeling. It is one of the most robust model for discrete
data modeling with an assumption that the dependent variable (number of
dimensions in which individual is poor) is distributed as Poisson and
its logarithm is a linear function if there are independent variables.
Wang and Famoya (1997) used this technique for the modeling of household
fertility decision, Femoya, et al. (2004) made an application of this
model on accidental data, John, et al. (2013) use this to assess
multidimensional poverty in Nigeria.
Poisson regression captures discrete and non-negative nature of
data, the second advantage of Poisson regression is that it allows
inference to be drawn on the probability of occurrence [Winkelmann and
Zimmermann (1995)]. Another important feature of Poisson regression is
that dependent variable is modeled as a deterministic function of
independent or explanatory variables, therefore randomness is
fundamental and not because of other factors.
The dependent variable in the model is the number of dimension in
which an individual is poor with non-negative numbers. An
individual's deprivation in different dimension is based on
different socio-economic and demographic characteristics, the expected
value of dependent variable (y) on a set of explanatory variables (x)
can be written as:
E(y/x) = [e.sup.(x'[beta])]
Where
Y = dependent variable
x = set of explanatory variables such as health, education, assets,
social wellbeing, environmental wellbeing, economic inclusion etc.
e = base of natural logarithm
x' = transpose of x
P = the vector of parameters.
The above equation shows that E(y/x) is greater than zero,
therefore an individual deprived in number of dimensions conditional on
x is the Poisson distributed with a probability of:
P(Y = 1/x) = [e.sup.-e(x'[beta])][e.sup.(x'[beta])]/y
Where y = 1,2, 3...8
The maximum likelihood poison multidimensional poverty equation can
be:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
X's in above equation are the set of independent
socio-economic variables which describe individual's
characteristics. The full model therefore can be written as:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
Where
[[delta].sub.i] = the expected number of dimension in which
individual is poor e = the base of natural logrithem [[beta].sub.0] =
the intercept [[beta].sub.j'] s = coefficient of regression
[x.sub.j]'s = explanatory variables
The study takes "a number of dimension in which an individual
is deprived" as dependent variable. To calculate the number of
dimension in which an individual can be poor, Alkair Foster (2010)
methodology has been used that helps to measure dimensions of poverty.
Furthermore Ataguba, et al. (2013) also used the same technique to find
dimensions in which an individual can be poor. Taseer and Zaman (2013)
used this technique to show time series breakdown in multidimensional
poverty in Pakistan. This methodology uses dual cut-offs to find
dimensions adjusted measure of poverty and is better than other
methodologies as it satisfying assumptions of monotonicity and
decomposability. To identify and measure multidimensional poverty,
head-counts and dimension adjusted head count rations are used. The
dimension adjusted head count Mo can be calculated as:
[M.sub.0] - [H.sub.0] x A
Where Ho is the proportion of people who are deprived in certain
dimension and A is the mean share of deprivation among the poor,
[M.sub.0] is used as a dependent variable in the model. Internationally
eleven dimensions has been selected to measure multidimensional poverty
among household or individuals but in case of M1CS dataset, it is only
useful to calculate seven dimensions. These include economic, housing,
air quality, health, education, water and sanitation, assets. Detailed
composition of these dimensions is given below in Table 3.
The contribution of human capital to poverty alleviation is proved
by previous literature. The development of human capital leads to an
increase in standard of living at household level. Communities with more
low-skilled workers in general are more likely to experience high rates
of poverty. The educational attainment as a measure of quality of human
capital is important, High educational attainment may imply a greater
set of employment opportunities that could decrease poverty [Cameron
(2000); Chaudhary, et al. (2009)]. The availability of education
facilities serve as a main indicator of remains poor. If the household
have an accessibility of school then there is a greater chance to get
rid from poverty. Theory shows a fundamental impact of health on
households, it is considered that the accessibility to health services
directly influence the productivity of individual household [McDonough,
et al. (2009); Zhong (2009)]. Another indicator of housing standards is
access to electricity. The housing indicators also affect the standard
of living of households. Employment is considered as an important factor
to affect poverty. The occupational affiliation of the head of household
is found to be an important determinant of poverty. The empirical
results suggested that the industry specific employment is necessary for
reducing poverty (increased per capita consumption and ultimately per
capita food consumption) [Sikander (2009)]. The employment trend is
defined by participation rate which is the ratio of the number of
workers to the number of adults in a household. The participation rate
is expected to be negatively correlated to poverty. Household income is
an important determinant of household expenditure since it serves as a
budget constraints to the amount that can be spent within a period,
there is also bound to be a correlation between income and poverty level
of a household, if all other things being equal. The household income is
also important to define the poor and non-poor households for further
analysis. In economic perspective, to judge the standard of living of
households, the household Property and Assets which contains the land,
livestock and other accessories of life also plays a role to determine
the poverty level among households.
As this paper is more concerned with relative poverty related with
socio-economic inclusion, capability etc., therefore this study also
uses some indices based on socioeconomic characteristics of individual
and household from where s/he belongs, developed by Wagle (2005). Since
integration of different theories would develop a realistic picture of
poverty, this study uses different dimension of poverty as explanatory
variable i.e. economic inclusion, social wellbeing, capabilities and
environmental wellbeing. The index of economic inclusion is developed
with the help of different variables that affect an individual in his
economic life. Theories suggest a strong link between employment type,
access to finance and occupation with standard of living [Athinkson
(1999); Wagle (2005)]. The employment in executive and professional
fields, employment in other fields, income, wealth and employment of
HHH's partner are some of indicators that are important for
economic inclusion of a person. Theory suggests that social wellbeing
can be measured by housing condition, electricity, access to safe
drinking water, access to secure housing tenure, type of toilet
facility, type of cooking fuel and type of assets etc. The index of
social wellbeing helps to predict the contribution of this index in the
poverty status of households. The approach to measure capabilities of
household to earn and to make its standard of living better proves to be
important for poverty analysis. Previous literature support that
educational achievement, health status, gender related discrimination
within households, household head health and access to health facilities
as indicators of capability. [Sen (1992); UNDP (2000); Wagle (2005);
Alkair (2007)]. Finally, the index of environmental wellbeing also plays
an important role in determining the level of poverty among urban
household, this index includes such indicators that directly affect the
health of household. The access of safe water, proper sanitation and
solid waste disposal facilities can be considered as some of the
important aspects of environmental conditions. All indices are
calculated with the help of principle component analysis.
VI. MARGINAL POPULATION IN VARIOUS POVERTY BANDS
The estimation of poverty line is very helpful to define various
bands of poverty such as extremely poor, ultra poor, non-poor etc.
[Economic Survey of Pakistan (2007)]. Population which consumes less
than 50 percent income of poverty line are categorised as extremely
poor, whereas population which lies between income groups (more than 25
percent of poverty line income) is considered as non-poor.
Around 70 percent of total population lives within extremely poor
and ultra-poor and only 6.8 percent of marginal class live out of
poverty in 2007 while in 2011 the poverty line is based on expenditure
approach, where 94 percent population appears to live in extreme
poverty.
VII. RESULTS AND DISCUSSION
The results from poison regression analysis is presented in Tables
5 and 6, the study use four models (two for each data set) to prove
hypothesis. Theory suggest a chain of marginality, social exclusion and
poverty, therefore model 1 of each dataset shows results that includes
marginality as an explanatory variable, while model 2 contains all other
variable of model one and use social exclusion index as an independent
variable to prove the theoretical link.
We found that coefficient has correct signs as defined in theory
with some minor contradictions. Results provided in Table 5 show that
income has a negative impact on the proportion of dimension in which
household can be poor and increase in income level will reduce poverty
threats by .02 percent ([e.sup.00002] = 1 .00), keeping all other
variables constant. The coefficient is significant at 1 percent. This
also proves the importance of multidimensional poverty that income has a
contributory role if defining a person poor but does not have a unique
role. While occupation of an individual also plays a negative impact on
the possibility to be poor and can draw him out from poverty, individual
who has good mean of earning than an individual with no or odd job has
lesser threat of poverty by 13 percent (e.0013=) at 1 percent level of
significance.
As far as the education of individual is concerned, compared to
those individuals who are illiterate, people having incomplete primary
education, threat of poverty is lower by 23 percent ([e.sup.0.2381] =
1.269), compared to not being literate, people having primary education
is found to be at minimal threat of poverty by 25 percent
([e.sup.0.2549] = 1.290) again assuming all other variables to be
constant. For those persons, who have matric and higher education have a
lesser threat to be poor by 29 percent ([e.sup.0.2926] = 1.3 3 9).
As far as housing condition is concerned, the variables reported
those individual who have poor housing condition, the result shows a
positive relationship of both variables, compared to people living in
better housing, the threat to be poor for those individual living in
poor housing increased by 11 percent(e 01103 = 1.116). The coefficient
of housing is significant at 1 percent.
An individual who has good amount of assets is also better off as
compared to individual with no assets; the possibility to be poor for
that individual is lower by 21 percent ([e.sup.0.2198] = 1.245) while
holding all other variable constant. Capability to be better off has
also strongly affect the status of poverty of an individual, a person
with good capabilities has a 31 percent (e0.3152) less chances to be in
multidimensional poverty than a person with no capabilities. Economic
inclusion also lower the risk of poverty, an individual who has greater
inclusion in economic activities has 55 percent ([e.sup.05578] = 1,746)
chances of deprivation in different dimension that a person with no
economic inclusion. Similar with social wellbeing, person with greater
social and civic services has less chances of deprivation than a person
with no social and civic services. The coefficient is significant at 1
percent level. Model 2 has almost same results with the same nature of
relationship.
Results of Poisson regression of 2011-12 data wave presented in
table 6 had only expenditure data while income aspect of household has
been ignored. Therefore the above table has two variables missing due to
non-availability of data, one is income of an individual and the other
is environmental wellbeing while one variable is additional i.e.
expenditures. According to results, expenditure has negatively affected
the risk to be in poverty, increase in expenditure will decrease
deprivation by .03 percent ([e.sup.0.0003] = 1.00), keeping all other
variable constant. The coefficient is significant at 5 percent. As far
as the education of individual is concerned, compared to those
individuals who are illiterate, people having incomplete primary
education, threat of poverty increase by 24 percent ([e.sup.0.2416] = i
269), compared to not being literate, people having middle level
education is found to be at a minimal threat of poverty by 3 percent
([e.sup.0.0333] = 1.034) again assuming all other variables constant.
For those persons, who have matric and higher education have a lesser
threat to be poor by 5 percent ([e.sup.0.0497] = 1.051). As far as the
housing condition is concerned, the variable reported those individual
who have poor housing condition, the results show a positive
relationship of both variable, compared to people living in better
housing, the threat to be poor for those individual living in poor
housing is increased by 5 percent([e.sup.0.0497] = 1.051). The
coefficient of housing is significant at 5 percent. Capability to be
better off has also strongly affect the status of poverty of an
individual, a person with good capabilities has a 17 percent
([e.sup.0.1732] = 1.189) less chances to be in multidimensional poverty
than a person with no capabilities. Economic inclusion also lower the
risk of poverty, an individual who has greater inclusion in economic
activities has 16 percent ([e.sup.0.1638 = 1.178) chances of deprivation
in a different dimension that a person with no economic inclusion.
Similarly with marginality index and social exclusion index, person with
higher marginality and social exclusion has high threat to be poor in
different dimensions than a person who is not at marginal position and
not socially excluded. Model 2 of this wave also shows similar results
with the same nature of relationship.
The results showing almost significant relationship with
relationship with poverty perceived in theory expect some of variable
which shows opposite results. Above results shows a picture of poverty
during two time period i.e. 2007-08 and 2011-12 respectively. Increase
in income and expenditure makes an individual better off and reduce
chances to be in poverty. An individual with high income and good
nutrition can access living facilities well and can be more productive
than a person with less food consumption [Headey (2008)]. Results also
show a negative and significant impact of income and expenditure on
deprivation and poverty in both waves. Wealth of an individual also
includes type and number of assets which an individual has, therefore
the state of poverty strongly depends upon the asset ownership of an
individual or household [Moser (1998, 2006)]. Results show a negative,
strong and significant relationship of assets ownership on risk of
multidimensional poverty, a person with good assets has lower chances to
be poor in different dimension than a person with no assets. Same
relationship is proved by Meek and Lansley (1985) and Milton (2003),
where lack of assets make a person more poor. Liverpool and Alex (2010)
shows a positive impact of asset building on consumption expenditure.
Another important determinant of poverty is education which is
proved by results from both waves. To make a detail analysis, we split
education into different levels and compare risk to be poor with
illiteracy. Higher education lower chances of poverty, Haroon (2009)
shows a positive impact of education on expenditures of household,
Dewilde (2004) proves that with increase in educational attainment, the
risk of poverty has been reduced, he tested this theory both on
uni-dimensional and multidimensional poverty risk, results also reflect
theoretical base, first wave supports the attainment in education lower
the chances for household to be poor, all results are significant at 1
percent level, while second wave (2011-12) shows a positive relation of
education attainment till primary level with poverty risk for an
individual, while education attainment (above primary and onward) will
lower risk of poverty significantly. Jhon, et al. (2013) also found a
positive impact of primary education attainment on multidimensional
poverty counts; similarly Dewilde (2004) also found greater proportion
of population with higher education within poverty. Narrayan, et al.
(2000) and Meek and Lansley (1985) also highlights assets, income and
education as important determinants of poverty.
Alkire (2008) gives high importance to housing condition as it
plays very important role in defining multidimensional poverty. If a
person living in poor housing, his chances of living in poverty is
greater, poor housing will reduce chances of having good living
facilities, relax life style and productive socio-economic contribution
[Taseer and Zaman (2013)]. Results support theory and prove a
significant impact of poor housing on chances to remain in poverty. As
far as health of an individual is concerned, in case of first wave,
results are insignificant and positive toward risk of poverty, while
second wave shows highly significant and positive relationship of poor
health of a person.
Sen (1992) approach of capability was defined as a transformed area
of poverty research, later OPH1 measure by Alkier and Foster (2008) also
focused on this dimension. She gave due importance to capabilities in
definition of multidimensional poverty measurement. Results show a
negative relationship between capabilities risk of poverty. The
improvement in capability will reduce the chances of poverty. Wagle
(2005) also proves a negative relation between poverty and capability
improvement. Similarly, the index of environment wellbeing has a
negative impact on poverty and environmental up-gradation ensures
betterment in standard of living of an individual. There is a
significant contribution of environment hazards in urban poverty,
degradation of renewable resources i.e. fresh water and poor mean of
waste management lead toward poor standard of living [Satterthwait
(2003)]. The index of social wellbeing also shows expected relation and
negatively affects the risk of poverty; an individual who is socially
empowered has fewer chances to be captured in the trap of poverty then a
socially deprived person.
The relationship between marginality and poverty shows different
results in two waves, during 2007-08, relationships show a negative
relationship, here one thing which should be considered while explaining
this relationship is that the marginality index is in form of least to
most marginal, the increase in value of marginality index shows high
marginalisation of specific individual, therefore if we assume
marginality leads to poverty then the expected relationship between
dependent variable and marginality index is positive which exist in
2001-12 data results. According to Franz, et al. (2011), it is not
necessary that a marginalised person is poor or a poor is marginalised,
however both are interlinked, therefore results support the theory
presented by Franzs, et al. (2011). In 2007-08, a negative relationship
between marginality index and poverty counts is reported, however in the
second wave relationship is positive and increase in marginality would
cause an increase in poverty counts. According to [Sen (2000)], the
concept of social exclusion is considered to be the root cause of
poverty. The composition of social exclusion index is similar to the
composition of marginality index, higher value shows higher exclusion or
vice versa. Both data results show a positive and significant
relationship between social exclusion and poverty counts. An increase in
social exclusion would cause ultimate poverty in different dimension of
socially excluded person.
VII. CONCLUSION
The study is an attempt to analyse the determinants of poverty
among marginalised population of urban Punjab. For this purpose, two
waves of Multiple Indicator Cluster survey (MICS) of the year 2007-08
and 2011-12 has been used. Among a sample set of more than two hundred
thousand, around 96000 were reported as marginalised based on
marginality index, and 33,629 were drawn as socially excluded from
marginal group, rest of the population is considered as marginal but not
socially excluded.
Results verify hypothesis and show that marginality is a root cause
of extreme poverty. As far as multidimensional poverty of this marginal
class is concerned, the population with no education or low level of
education is highly poor in multidimensional way, the extent of economic
inclusion, social betterment, capability improvement also positively
affect an individual and his poverty count is reduced with the
betterment of above mentioned indicators. Overall results confirm the
research question that marginality cause poverty.
On the determinants side, the income support programs cannot break
the vicious circle of poverty until and unless policies focus is toward
the determinants of poverty. Along with education, health and housing,
the role of capabilities and environment, exclusion and marginality
cannot be ignored. There is a strong link of these variables with
poverty. Therefore a need to focus on these determinants is important
and providing income, skills, education and other related factors should
be the focus of any policy decision. Marginality and social exclusion
may be a new concept for poverty reduction policy-making, but we
can't deny this fact that in Pakistan, the focus of poverty
reduction strategies is on curative measures, not on preventive
measures. Therefore those factors which exclude a household or
individual from community is not at target of policy makers.
Countries are going to treat marginal communities separately to
make specific policies for their benefits but in Pakistan still there is
dearth of literature and no dataset exist which could cover issues of
marginal community specifically. A need to study marginal class and
their problem through survey and research is strongly needed. Moreover,
a new diverted focus of policy should be on the treatment of poverty
among disadvantaged class. The separate survey to target marginalised
communities if not possible then at least inclusion of this dimension in
some national and provincial level data set can fulfil the purpose.
These people have different nature of issues in their social, economic
and cultural lives which are related with socioeconomic service delivery
mechanism together with infrastructure provision.
APPENDIX
[ILLUSTRATION OMITTED]
Kanwal Zahra <Kanwalzahral@yahoo.com> is PhD Scholar, GC
University, Lahore and Assistant Professor, UCP Business School,
University of Central Punjab, Lahore, Tasneem Zafar
<Incharge.eco@gcu.edu.pk> is Assistant Professor, Department of
Economics, GC University, Lahore.
REFERENCES
Alkire, S. (2008). Choosing Dimensions: The Capability Approach and
Multidimensional Poverty. SSRN Journal.
Angemi, D. (n.d.) Measuring Household Vulnerability in the Context
of Poverty Education: Evidence from Uganda. SSRN Journal.
Aya K., A. (2009) Social Exclusion and Earlier Disadvantages: An
Empirical Study of Poverty and Social Exclusion in Japan. Social Science
Japan Journal 13:1, 5-30. Calvo, C. (2008) Vulnerability to
Multidimensional Poverty: Peru, 1998-2002. World Development 36:6,
1011-1020.
Dewilde, C. (2003) A Life-Course Perspective on Social Exclusion
and Poverty. British Journal of Sociology 54:1, 109-128.
Headey, B. (2008) Poverty Is Low Consumption and Low Wealth, Not
Just Low Income. Soc Indie Res 89:1, 23-39.
Labar, K. and F. Bresson (2011) A Multidimensional Analysis of
Poverty in China from 1991 to 2006. China Economic Review 22.
Lyberaki, A., and PI. Tinios (2005) Poverty and Social Exclusion: A
New Approach to an Old Issue. In A. Borsch-Supan, A. Brugiavini, H.
Jurges, J. Mackenbach, J.Siegrist and G. Weber Health Ageing and
Retirement in Europe: First Results from the Survey of Health, Ageing
and Retirement in Europe. Mannheim, pp. 302-309.
Liverpool-Tasie, L. and A. Winter-Nelson (2010) Asset versus
Consumption Poverty and Poverty Dynamics in Rural Ethiopia. Agricultural
Economics 42:2, 221-233.
Mitra, A. (2004) Informal Sector, Networks and Intra-City
Variations in Activities: Findings From Delhi Slums. Rev Urban & Reg
Dev Stud 16:2, 154-169.
Moser, C. (1996) Household Responses to Poverty and Vulnerability.
Washington, D.C.: Published for the Urban Management Programme by World
Bank.
Moser, C. (1998) The Asset Vulnerability Framework: Reassessing
Urban Poverty Reduction Strategies. World Development 26:1, 1-19.
Moser, C. (n.d.) Asset-Based Approaches to Poverty Reduction in a
Globalised Context. SSRN Journal.
Pakistan Planning Commission (2011) Change in the Cost of Food
Basket, Annual Report, Nutrition Section, Planning and Development
Division, Government of Pakistan.
Ruth, L., P. Christina, E. Fahmy, D. Gorden, L. Eva, and D. Patsios
(2007) The Multidimensional Analysis of Social Exclusion. University of
Bristol.
Sen, A. K. (1997) On Economic Inequality. New York: Clarendon
Oxford Press.
Stefan, S. (2004) Urban Vulnerability in Afghanistan: Case Studies
from Three Cities. The Afghanistan Research and Evaluation Unit, Working
Paper Series.
Stefan, D. (2007) Globalisation and Marginalisation in Africa:
Poverty, Risk, and Vulnerability in Rural Ethiopia. World Institute for
Development Economic Research. Working Papers UNU-WIDER Research Paper.
Stefan, D., J. Hoddinott, P. Krishnan, and T. Woldehannna (2008),
Collective Action and Vulnerability: Burial Societies in Rural Ethiopia
(CAPRi Working Paper No. 83).
Susan, L. and O. W. Mittal (2001) Urban Poverty and Vulnerability
in India, DEFID.
Thorat, S., P. Attewell, and F. Rizvi (2009) Urban Labour Market
Discrimination. Indian Institute of Dalit Studies, New Delhi. (Working
Paper Series 3:1).
Thorat, S. and N. Sabharwal (2010) Caste and Social Exclusion,
Issues Related to Concept, Indicators and Measurement. IIDS and UNICEF
Working Paper Series, 1: 2.
United Nations (2007) Literature Review on Social Exclusion in the
ESCWA Region. Economic and Social Commission for Western Asia (ESCWA).
New York: United Nations.
Vijaya, M., R. L. Rahul, and Hema (2014) Moving from the Household
to the Individual: Multidimensional Poverty Analysis. World Development
59.
Wagle, U. (2005) Multidimensional Poverty Measurement with Economic
Well-being, Capability, and Social Inclusion: A Case from Kathmandu,
Nepal. Journal of Human Development 6:3, 301-328.
Walter, P., C. Susan, and H. Frank (2007) Benchmark Analysis for
Quantifying Urban Vulnerability to Terrorist Incidents. Risk Analysis
27: 6.
Whelan, C. and B. Matre (2005) Vulnerability and Multiple
Deprivation Perspectives on Economic Exclusion in Europe: A Latent Class
Analysis. European Societies 7:3, 423-450.
Zahra, K., and T. Zafar (2014) Marginality and Social Exclusion: An
Interactive Approach. (GCU Working Paper).
Comments
* The paper aims to provide a district level analysis which focus
on micro level poverty analysis of the marginalised people of Punjab
with focus on whether these people are living in extreme poverty or out
of poverty. Paper also find out the determinants of MPI by using Poisson
regression The study has defined the marginality as a state situated at
the margin, this could lead toward social exclusion hence poverty or a
marginal person can be out of poverty. Please rephrase the definition of
marginality because it is missing the social-economic aspects.
* I not found much difference between theoretical framework and
literature review. Author has given a uni-directional framework moving
from marginalisation-social exclusion--MPI. I think it is not
unidirectional. Current deprivation can also transmit social exclusion
in future and especially look at the socially excluded people, they are
transferring their current deprivations into their children by making
them potential future social excluded people. You also mentioned similar
statement at pp. 8 that marginality has casual relation with poverty.
* At pp. 9 author give powerful statement that "urban areas of
Punjab remain in extreme poverty". I think its cannot be justified.
Though there are pukets of rural poverty with a lot of regional
variation but you can find comparatively less urban poverty differential
even north, central and south. You have skipped three studies which have
applied the small geographical technique on two micro data-sets to
compare poverty levels across more than one hundred districts of
Pakistan. Jamal (2007) and Cheema (2010) utilised the HIES 2004-05 and
PSLM 2004-05 for the district level poverty comparison while Ali (2011)
has applied this technique on the 2007-08 HIES 2007-08 and 2007-08 MICS
(Punjab) to predict poverty at district and tehsil levels for Punjab.
* On pp. 9-10 please correct the sample information of MICS
2011-12. 102,545 were conducted. The study has taken the concept of
marginalised and socially excluded from Zahra and Tasneem (2014). I not
found the study in reference list so unable to know how concept has been
measured. Though annexure 1 and 2 has explained the marginality and
exclusion but definition is not clear, i.e. in marginality people living
along not in youth (16+) but youth is also comprises of 16+, second how
you tackle the students as your unit of analysis is individual but I
don't know what age is in unit of analysis, then majority of the
characteristics coming from hh, so it should be defined at hh level.
Then almost the variables in marginality and social exclusion are same
i.e. economically inactive/unemployed in marginal and employment in
social exclusion, education vs literacy or low educational attainment,
electricity/gas/rented home vs freezer/oven/computer, average number of
persons per room vs housing congestion. Same are the indicators of MPI
so if you put all the three annexure tables together, you cannot found
any major difference. Then if you see annexure Table 3, you are taking
majority of indicators at hh level so why then analysis at individual
level. The study has taken 7 dimensions and a lot of indicators (though
definition is not given of indicators.) need to check
multi-collinearity, and robustness.
* At pp. 12 author stated that the study takes "number of
dimension in which an individual is deprived" as dependent
variable.
* In Table 4.1, there is need to explain poverty line i.e. author
reported 1668 for 2011 but it is not. Second how poverty is estimated
from MICS data, including technique and poverty line. Third you cannot
compare income based vs consumption base and you can find a lot of
difference in Table 4.1.
* I am concerned on the theoretic building of poisson regression
model. Because majority of your variables are either facing
multicollinary or they also exist in the estimation of dependent
variable, (what you do if income is 0 as explanatory variable. Education
is itself determinant of income (Table 4.2). Occupation and income,
assets and income and then social, economic well-being and marginally
also capture the same.
* In Table 4.3 food expenditure is itself determined by MPI.
It's totally wrong
* In annexure GIS mapping, deprivation is 0 or high in Gujrat,
Rawalpindi, Attock is unable to understand, the range is also
questionable. All these numbers make the whole paper ambiguous.
Shujaat Farooq
Pakistan Institute of Development Economics, Islamabad.
(1) UN (2007) quoted in Oxfam (2009)
(2) Oxfam (2009).
(3) Planning Commission of Pakistan (2011)
Table 1
Dimension and Indicators of Marginality in Punjab
Dimension Indicators
Residential People living alone, people who are not in youth (aged
Instability 16+), average number of persons per room, dwelling that
are apartment and small building, population of
divorces/widowed, people living on rent, people without
access to electricity and gas
Material People aged 20 and above without secondary school,
Deprivation people receiving transfer payments, unemployed
population (aged 15+, population in low income
quintile, type of housing, unpaid family worker)
Dependency Dependency ratio, female population, population aged
65+, population (aged 15+) economically inactive.
Ethnicity Language, Religion
Reproduced from Zahra and Tasneem (2014), Marginality and Social
Exclusion in Urban Punjab: A Spatial Analysis, Working Paper, GC
University, Lahore.
Table 2
Dimension and Indicators of Exclusion in Urban Punjab
Economic Employment (unemployed or discouraged employment),
Exclusion Population at poverty risk, Material deprivation of
housing, housing congestion
lack of washing machine, freezer and oven, lack of
computer and internet accessibility, lack of access of
bank account, overcrowding
Exclusion Low educational achievement (basic schooling) and early
from Social school leavers, No access to water and sanitation
Services facility, Household with young children not in school,
Cannot read or write, Poor general health, Poor physical
health, Disable child quality
Cultural low neighbourhood, membership of community centre etc.
Exclusion little social support
Reproduced from Zahra and Tasneem (2014), Marginality and Social
Exclusion in Urban Punjab: A Spatial Analysis, Working Paper, GC
University, Lahore.
Multidimensional poverty has been evaluated in extracted population
sample.
Table 3
Dimensions and Indicators used for Dependent Variables
Dimensions Indicators
Living Standard Housing type (floor, roof and walls material),
dwelling type, rooms congestion, electricity,
Environment Type of Fuel, open dumps nearby, solid waste
disposal, kitchen for cooking
Assets TV, Refrigerator, AC, Vehicle (car, Motor Cycle),
Oven, Washing Machine, Room cooler
Education Education of HHH, education attainment (primary)
Health Vaccination, Disable HH member
Livelihood Employment type, other source of income
Water and Proper means of water, proper mean of sewerage,
Sanitation toilet facility, HH use boiled/filtered water for
drinking purposes
Table 4
Marginal Population in Various Poverty Bands
2007-08 2011-12
Income based Food Consumption
Poverty Line based Poverty
($1.5 per Day) (Rs 1668) (3)
Extremely Poor > 50% 40.8 94
Ultra Poor 50% <X>75% 30.0 4.1
Poor 75% <X>100% 11.0 1.2
Vulnerable 100% <X>125% 5.8 0.3
Non-poor 125% <X 6.8 0.4
Planning Commission of Pakistan (2011).
Table 5
Poisson Regression Output (2007-08)
Dependent Variable: Poverty Counts
Model 1
Coefficient Standard Error
Income -.000261 *** 0.000157
Poor Health .00362 .01121
No Education (reference)
Pre-primary -.238136 *** .019691
Primary -.254944 *** .008643
Middle -.26673 *** .0107994
Matric -.292634 *** .022817
Higher -.292006 *** .0278123
Madrassa -.316337 *** .0978124
Poor Housing Condition .11037 *** .00729
Occupation -.00135 *** .000102
Assets -.21986 ** .09553
Capability -.315201 *** .035955
Environment Wellbeing .26406 *** .03086
Social Wellbeing -.71456 *** .096468
Economic Wellbeing -.55788 *** .072224
Marginality Index -.03103 ** .007036
Social Exclusion Index -- --
Log Likelihood -50127.518
Pseudo R2 .0239
LR [chi square] (12) 2457.01
Prob> [chi square] 0.0000
Model 2
Coefficient Standard Error
Income -.000259 *** .000157
Poor Health .001378 .011224
No Education (reference)
Pre-primary -.22552 *** .01975
Primary -.23253 *** .009776
Middle -.237417 *** .012318
Matric -.25584 *** .023737
Higher -.25359 *** .028669
Madrassa -.28196 *** .098035
Poor Housing Condition .11064 *** .00729
Occupation -.00130 .000104
Assets -..19139 ** ..095745
Capability -.26964 *** .036493
Environment Wellbeing .26964 *** .03087
Social Wellbeing -.724314 *** .096684
Economic Wellbeing -..632038 *** .070282
Marginality Index -- --
Social Exclusion Index .02941 .00722
Log Likelihood -50129
Pseudo R2 .0239
LR [chi square] (12) 2454.04
Prob> [chi square] 0.0000
Table 6
Poisson Regression Output (2011-12)
Dependent Variable: Poverty Counts
Model 1
Coefficient Standard Error
Food Exp -.00035 ** .000151
Poor Health .28101 *** .050004
No Education (reference)
Pre-primary .241633 *** .05386
Primary .225908 *** .00840
Middle -.03337 ** .01031
Matric -.045371 *** .00970
Higher -.04938 *** .00677
Poor Housing Condition .04979 ** .016667
Occupation .000184 .000239
Assets -2.3437 *** .161364
Capability -.173256 ** .069413
Social Wellbeing 3.6445 *** .12478
Economic Inclusion -.163873 *** .022886
Marginality Index .035296 *** .006772
Social Exclusion Index -- --
Log Likelihood -58456
Pseudo R2 .019
LR [chi square] (I2) 2275.35
Prob [chi square] x2 0.0000
Model 2
Coefficient Standard Error
Food Exp .000346 ** .000513
Poor Health .26754 *** .04996
No Education (reference)
Pre-primary .244616 *** .053865
Primary .231517 *** .008779
Middle -.03517 ** .011005
Matric -.028226 ** .010339
Higher -.028919 ** .01123
Poor Housing Condition .03649 ** .01650
Occupation .000205 .000239
Assets -2.4809 *** .16043
Capability -.151884 ** .06940
Social Wellbeing 3.7985 *** .12019
Economic Inclusion -.10671 *** .01974
Marginality Index -- --
Social Exclusion Index .01079 ** .00513
Log Likelihood -58468.287
Pseudo R2 .0189
LR [chi square] (I2) 2252.63
Prob [chi square] x2 0.0000
