Measuring multidimensional poverty and inequality in Pakistan.
Sial, Maqbool H. ; Noreen, Asma ; Awan, Rehmat Ullah 等
Using household level data, the study measures multidimensional
poverty and inequality in Pakistan. The indicators used in the analysis
are equivalent per adult expenditures, maximum years of schooling,
enrolment, immunisation against measles, post-natal care, crowding, gas,
electricity, safe drinking water and sanitation facility. The results
show that poverty and inequality in uni-dimensional as well as
multidimensional context has declined in Pakistan between 2005-06 and
2010-11. The major contributing factors in poverty are education and
health. The dimensional disaggregation of poverty also showed that
crowding and access to safe drinking water has greater share in poverty.
The study also finds that inequality has declined more in monetary terms
as compared to multidimensional measure of inequality for Pakistan.
Moreover, it was found that health inequality is on the rise in
Pakistan.
JEL Classification: 132, P36
Keywords: Multidimensional Poverty Index (MPI), Multidimensional
Gini Index (MGI), Millennium Development Goals (MDGs), Deprivation,
Pakistan
1. INTRODUCTION
The key development objective of Pakistan, since its existence, has
been to reduce poverty, inequality and to improve the condition of its
people. While this goal seems very important in itself yet is also
necessary for the eradication of other social, political and economic
problems. The objective to eradicate poverty has remained same but
methodology to analysing this has changed. It can be said that failure
of most of the poverty strategies is due to lack of clear choice of
poverty definition. A sound development policy including poverty
alleviation hinges upon accurate and well-defined measurements of
multidimensional socio-economic characteristics which reflect the ground
realities confronting the poor and down trodden rather than using some
abstract/income based criteria for poverty measurement. Conventionally
welfare has generally been measured using income or expenditures
criteria. Similarly, in Pakistan poverty has been measured mostly in
uni-dimension, income or expenditures variables. However, recent
literature on poverty has pointed out some drawbacks in measuring
uni-dimensional poverty in terms of money. It is argued that
uni-dimensional poverty measures are insufficient to understand the
wellbeing of individuals. Poverty is a multidimensional concept rather
than a unidimensional. Uni-dimensional poverty is unable to capture a
true picture of poverty because poverty is more than income deprivation.
Multidimensional wellbeing is not a new concept. Sen (1976) is
among the pioneers to conceptualise the wellbeing in an alternate and
direct way using the concept of capability and functioning. Income is
rather an indirect approach to evaluate individual welfare since it is
required to purchase basic needs for example food, shelter, clothing,
etc., assuming a well-functioning market. Consequently, it is true to
say that to acquire certain goods and services income is needed. So,
poverty is not the state of deprivation of certain level of income, that
is, one dollar per day or two dollars per day but poverty is a state of
multiple deprivations that poor faces. This multiplicity of dimensions
leads to a broad definition of poverty. Income data is not correctly
available and also there are difficulties in adjusting data for prices
and inflation. Multidimensional poverty does not fluctuate due to
inflation. So, this is a relatively stable measure. Indicators reflect a
relatively long term accumulation of welfare.
Many studies in 1990s have put a great emphasis on the relationship
between economic growth, inequality and poverty. Inequality is also a
multidimensional phenomenon. Only income or consumption inequality does
not give true snapshot of inequality. It has considered that the
combination of economic growth and inequality reduction policies are the
key determinants of poverty reduction. Eradication of poverty is one of
the most important objectives of all countries but the question arise,
how to tackle this ambiguous problem? Although measuring MPI is a
difficult task but due to the importance of MPI some developed and
developing like Mexico, Colombia, Philippines, China, Brazil, Bhutan,
Malaysia, Indonesia, Chile etc. have adopted multi-dimensional poverty
estimation. Many researchers have investigated MPI for different
countries like Metha and Shah (2003) in India, Justino (2005) in Brazil,
Batana (2008) in Sub-Saharan countries, and Alkire and Santos (2010) in
America. Computation of Multidimensional inequality is an exigent
exercise as many variables contribute in it. Multidimensional inequality
has been investigated in different countries like Justino (2005)
examined multidimensional poverty in Brazil, Decancq, et al. (2009) in
several countries, Aristei (2011) in Italy, Decancq and Lugo (2012) in
Russia and Rohde and Guest (2013) in US.
In Pakistan, for alleviation of poverty, different social programs
like Benazir Income Support Program (BISP) and Wasila-e-Rozgar scheme
etc., have been introduced. Implementation of Poverty Reduction Strategy
Papers (PRSP) approach and pledge to achieve Millennium Development
Goals (MDGs) by Government of Pakistan reveal the importance of and need
of poverty reduction. All poverty reduction strategies and social safety
nets require the accurate analysis of poverty estimates to reach the
primal objective of these programs. For achieving eight Millennium
Development Goals, it is necessary to gauge the multidimensionality of
poverty and inequality.
Main objective of this study is to measure multidimensional poverty
and multidimensional inequality in Pakistan. Rest of the study is
organised as Section 2 illustrates data and methodology. Results,
conclusion and policy recommendations are discussed in Section 3.
2. DATA AND METHODOLOGY
This study employs Household Integrated Economic Survey (HIES)
(2005-06 and 2010-11) for analysis of Multidimensional Poverty and
Inequality in Pakistan. The study follows Alkire and Foster (2007)
methodology for measuring Multidimensional poverty and Gini index for
Inequality. It is easy to compute and interpret, intuitive and fulfils
many desirable axioms. This is also called Adjusted Headcount Ratio
(Mo>. Mo is applicable when any one of dimensions is ordinal,
numerical values have no meaning, for example, qualitative in nature.
Alkire and Foster Methodology (AFM) does not assume that data is
continuous. Brief and simple steps involved in computation of MPI in
mathematical as well as non-mathematical form are given below:
Step one is choice of unit of analysis. Unit of analysis may be
household, individual, community, district or country. Choice of unit of
analysis depends on research question. Second step is choice of
dimensions. Dimension may be selected due to public consensus, empirical
evidence or convenience due to data availability. Choosing indicator for
each dimension is third step. For more accuracy and parsimony, it is
necessary to use few reasonable indicators. Fourth step is to apply
poverty lines for each indicator and convert achievement matrix into
deprivational matrix. Replace individual's achievement with 0 or
one. Intuitively Equation 1 shows that if achievement of nd*household in
dimension D is less than dimensional cut-off za then that household is
declared as poor and vice versa.
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)
Next step is to count the number of deprivations for each
individual and get one column vector. For simplicity, equal weights are
assumed to apply. Then set second cut off for identification, k is
applied. If the number of deprived dimension is smaller than k then the
household is non-poor and vice versa according to identification
function. Identification function pk is defined in Equation 2.
[[rho].sub.k] = ([Q.sub.n], z) = 1, if [S.sub.n] [greater than or
equal to] k (2)
Now a censor matrix is defined, having zeros and ones for non-poor
and poor households respectively. As AF methodology is concerned, MPI is
calculated by the product of two components. First component of MPI is
Multidimensional Headcount Ratio and second is Average Intensity of
Poverty. So calculate headcount ratio (H) by dividing the number of poor
by total population. Mathematically M = h (Q; z)is defined in Equation
3.
H = q/N (3)
where
N = Total population
q = q (Q; z) = [[summation].sup.N.sub.n=1] [[rho].sub.k]([Q.sub.n],
z) in set of [z.sub.k] = {n : [[rho].sub.k]([Q.sub.n*], z) = 1}
Average intensity (A) is ratio of sum of deprivation of each
individual to total number of poor. It gives the fraction of dimensions
in which average household experienced multidimensional poverty. MPI is
simple product of the average share of deprivation with raw headcount
ratio, which is given in Equation 4.
[M.sup.0] = A x W ... (4)
MPI measures can also be decomposed by dimensions as given in
Equation 5.
Dimensional Contribution = [mu]([g.sup.[alpha].sub.d](k))/M(Q;z)
(5)
2.1. Dimensions, Weights and Cut-offs
For accurate measurement, selection of appropriate dimensions is as
important as accurate data. It is an important and most challenging
task. But for this, there is no hard and fast rule. According to Sen,
appropriate dimensions are those which have importance for individuals
and society, and have importance regarding policy making. Based upon
data availability regarding MDGs, this study has selected four
dimensions, Education, Health, Expenditures and Living Standard, with
ten indicators. Eight out of ten indicators are related to MDGs.
MPI can be measured at individual or household level through AF
methodology, but if we select individual as a unit of analysis there
might be a problem of data availability. In Micro survey expenditures,
detail is present only on household level so it is most appropriate to
analyse poverty on household level. In this study unit of analysis is
household level. All four dimensions are equally weighted, equal to one
forth or 25 percent. Moreover, all indicators in all the dimensions are
also equally weighted.
2.1.1. Education
Education enables an individual, hence a nation to adopt new
technology and technological changes. Income also depends on complete
years of schooling. If a child is not attending school, it lessens the
chances of increase in future income of a household. That's why
education is the second goal of MDGs. (1) So it is essential to dissect
education poor areas and take some measures. In line with MDG two, this
study has used five years of education and school enrolment of schooling
children as indicators of education. Current education level shows level
of knowledge of a household because any one of educated member has
positive externalities on other members of the household [Basu and
Foster (1998)]. So the household is referred as deprived if no member
has completed five years of education. The current enrolment status of
all school going age members shows an increase in present and future
abilities. The household is declared as poor if any school going age
child is not enrolled in school.
2.1.2. Health
Health like education is also a key variable of development. A
person with good health can earn enough income to meet the basic
necessities of his household. That's why MDGs put greater emphases
on health. Three out of eight MDGs are in one or other way related to
health. This study, based on data availability, has included
immunisation against measles and postnatal care as indicators of health
which are directly linked to fourth and fifth goal of MDGs respectively.
Household is declared as deprived in immunisation if no child is
immunised against measles. In Postnatal care Household is referred as
deprived if mother of a family never goes for postnatal check-up.
2.1.3. Expenditures
As multidimensional poverty emphasises on other dimensions of
poverty, it does not mean to ignore monetary dimension. Monetary
dimension is as important as others because income provides power to
purchase other basic necessities. Many empirical studies excluded
income/expenditures due to lack of data on income and expenditures.
However, PSLM survey has data on all dimensions. There are three
possible choices for material wellbeing income, asset ownership and
expenditures. People do not report their true income level. They report
low income level than actual may be due to fear of theft or income tax.
Furthermore, income of agriculture and self-employed is difficult to
estimate. Second indicator, asset ownership is supposed to be a good
measure because it is free of these issues but the choice of a set of
assets is highly crucial and questionable. Expenditure data is
relatively credible than income data.. So this study has used
expenditure data as a proxy of monetary wellbeing.
2.1.4. Living Standard
A single variable is not able to measure living standard. Due to
limitation of data this study has selected only five indicators for
living standard; access to safe drinking water, electricity, gas,
sanitation and crowding per room. These indicators directly or
indirectly are included in MDGs.
Life is impossible in the absence of water. But contaminated and
grimy water grounds many diseases like diarrhoea and hepatitis leading
to many deaths in Pakistan. Safe drinking water is related to seventh
MDG.A household is deprived in water dimension if the household has no
access to drinking water or more than 30 minutes consume to reach the
source of safe drinking water.
Access to improved sanitation is a vital need for dignity and
health of people. Sanitation is directly related to hygienic problems.
For safety of health access to improved sanitation is essential. It is
also related to seventh. A household is declared as poor if the
household does not have an improved toilet.
Electricity is an important indicator of living standard.
Electronic devices can be used with electricity and these electronic
products enhance productivity and leisure. A household is deprived in
this dimension if it has no electricity connection.
Gas is also indirectly related to MDGs; Bio mass fuel causes
environmental degradation. So indirectly gas provides environmental
sustainability and is also linked with 7th MDG. A household is
pronounced as poor if it has no gas connection.
Crowding symbolises the number of persons who share a room for
sleeping. More persons sharing one room indicate the shortage of
facilities hence a low living standard. Indirectly this indicator is
related to MDG five. A household is deprived if more than three persons
are sharing one room for sleeping.
2.2. Multidimensional Inequality
Fisher (1956) is one of the pioneers of Multidimensional
Inequality, who have given the idea of Multidimensional matrix. Kolm
(1977), Atkinson and Bourguignon (1982), Walzer (1983), Gajdos and
Weymark (2005) and Decancq and Lugo (2011) have proposed
Multidimensional Inequality indices. Decancq and Lugo (2012) aggregate
in reverse direction, first aggregate across dimensions and then across
individuals, this technique takes into account correlation between
dimensions. So, this study has employed Decancq and Lugo
Multidimensional methodology for measuring Multidimensional Gini
coefficient.
For measuring Multidimensional inequality Gini coefficient
distributional matrix can be compared with social welfare function.
Distributional matrix is similar as achievement matrix in MPI. Social
welfare function can be constructed by double aggregation. First
aggregated across dimensions to get individual welfare function and then
aggregated across individuals for social welfare function. Same or
different weights can be assigned to each dimension. Weights reflect the
relative importance of a dimension. Weights must be assigned in such a
way that the sum of all weights must be equal to one, mathematically
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII].
Social wellbeing function can be written as:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]
[W.sub.d] = weights assigned to each dimension
[r.sup.n] = rank of individual n on the basis of S
[delta] = inequality inversion parameter
[beta] = substitution parameter
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7)
So [beta] mirrors the degree of substitutability among the
dimensions of wellbeing. Specifically, [beta] is related to the
elasticity of the substitution c between the dimensions, and equals 1-1/
[sigma]. [beta]=1, represents perfect substitutability between the
dimensions of wellbeing and [beta] = -[infinity], denotes dimensions are
perfect complementry; at the extreme, individuals are judged by their
worst outcomes.
Multidimensional Gini index can be defined as the fraction of an
aggregated amount of dimensions of a distributional matrix by aggregated
amount of dimensions if dimensions are equally distributed. It is
analogous to Uni-dimensional Gini index which is also ratio of
aggregated gap of income to the equally distributed income. So
Multidimensional Gini is defined as below:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (8)
The value of Gini co-efficient lies between zero and one
(0<Gini<l). Gini coefficient equal to zero shows a perfectly equal
distribution of income (every person has the same level of income) while
equal to one shows a perfectly unequal distribution of income (some
persons have more income and some have less income).
Choice of different dimensions for multidimensional inequality is a
crucial task and can be chosen with various justifications. Following
dimensions are used for Multidimensional Inequality: Monetary wellbeing
(consumption expenditures), Educational achievement and Health. This
study has utilised maximum years of education attained by any member of
a household as a proxy of educational wellbeing.
There are many proxies for Health like BMI, stunning and wasting
for children, but these are not monotonically increasing function of
health. Alternate to this, health risk index, proposed by Seth (2010),
is a better measure of health quality. This index captures the risk of
health of a household. The higher the value of health index mirrors the
better quality of health. Health risk index combines three
sub-indicators. The first indicator, safe drinking water Iw=1 if a
household has access to safe drinking water and Iw=0, otherwise. The
second indicator is access to improved toilet, It=l if a household has
access to improved sanitation and It=0, otherwise. Last indicator,
crowding Ic=Mn/Rn, where Mn is number of members and Rn is number of
rooms. It is difficult to decide normatively which indicator has more
importance for health risk index so equal weights have applied. Health
risk index for N household is defined in Equation 9.
H.R.I = 1/3 [I.sub.w] + 1/3 [I.sub.t] + 1/3 [I.sub.c] (9)
This equation shows simple Health Risk Index with equal weights and
ordinal data. As these dimensions have only value judgment and all
researchers may not agree upon any weighting scheme, so equal weights is
a better option.
3. RESULTS AND DISCUSSION
3.1. Results of Multidimensional Poverty
Income or expenditures are not the only determinant of poverty,
many factors other than income/expenditures also contribute in poverty
like poor health, lack of education, poor housing and low living
standard. Poverty is something beyond income deprivation. This study has
estimated MPI at National level.
This study applied Alkire and Foster (2007) methodology by using
four dimensions; health, education, expenditures and living standard for
measuring multidimensional poverty. Figure 1 shows the percentage of
multidimensional deprived people by using ten indicators under the
umbrella of four dimensions and by using equal weights. In 2005-06, 51
percent people were multidimensional deprived where as 49 percent people
were multidimensional non-deprived.
[FIGURE 1 OMITTED]
In 2010-11 multidimensional deprived population was 35.86 percent
while the rest were non-deprived. Multidimensional poverty has declined
by 15 percent during 2005-06 to 2010-11.
3.2. Dimensional Share to overall MPI
From Figure 2, it is clear that education contributes 23 percent in
2005-06 and 28 percent in 2010-11 to overall poverty, Health contributes
32 percent and 28 percent in 2005-06 and 2010-11 respectively while
living Standard contribution is 25 percent and 28 percentage
respectively. Expenditure share to overall poverty is 20 percent and 16
percent in 2005-06 and 2010-11. Contribution of education and living
standard has increased over time. The reason for increase in share to
poverty was due to over population, decline in quality of water and
increase in education inequality. Thus these findings conclude that to
curb multidimensional deprivation, resources must be allocated according
to the percentage of the share of each dimension in overall poverty.
Dimensional contribution can be more zoomed by considering
percentage deprivation in each indicator. Figure 3 gives a more detailed
picture of multidimensional poverty through percentage of deprived
population in each indicator.
Above diagram illustrates that during 2005-06 and 2010-11, about 77
percent and 28 percent households are deprived in post natal care, 25.5
percent and 19.3 percent households in immunisation against measles,
71.7 percent and 63 percent households have not a gas connection
respectively. It is also depicted that the households have small houses
as compared to their family, i.e. 45.8 percent and 53.7 percent
households share a room more than three persons in 2005-06 and 2010-11.
Congestion may cause many diseases which spread through respiration. In
2005-06 and 2010-11, 45.3 percent and 42 percent of households had no
access to an improved toilet. In Pakistan, in 2005-06 and 2010-11, 15.3
percent and 16.2 percent households had no access to safe drinking water
or it may consume 30 minutes to reach at the source of drinking. 14.3
percent and 9.4 percent households have no electricity connection as a
lightning source while 22 percent of the population was living below
Poverty line (BPL) in 2005-06 and 2010-11.
According to Union approach adjusted headcount ratio is 36.83
percent and 25.23 percent whereas Intersection approach shows zero
percent headcount ratio in 2005-06 and 2010-11. Regarding Dual Cut off
approach, poverty cut off "K" must be lie between two
approaches. At "K=2" headcount ratio is 33.74 percent and
22.89 percent. As "K" increases headcount ratio decreases such
as at k equal to three, four, five, six, seven and eight headcount ratio
is 28.36 percent, 23.11 percent, 18.86 percent, 12.92 percent, 8.87
percent, 4.97 percent and 1.58 percent respectively in 2005-06. In
2010-11 for k equal to three, four, five, six, seven and eight headcount
ratio is 18.13 percent, 13.26 percent, 9.92 percent. 6.29 percent, 3.10
percent, 1.41 percent and 0.1 percent respectively. The estimates of
headcount ratio illustrate that multidimensional poverty has declined at
all cut offs during 2005-06 to 2010-11.
3.3. Results of Inequality in Pakistan
This study analysed inequality in wellbeing in Pakistan for 2005-06
and 2010-11. Data used in this section have been also drawn from PSLM
conducted by Pakistan Bureau of Statistics. In addition to consumption
expenditure, facts of health and education have also been incorporated.
Equivalent per adult expenditures, proposed by Planning Commission of
Pakistan, are used. In analysis, maximum years of education of any of
the family member and Health Risk Index are included.
For computation of Gini coefficient, this research used four
fundamental decisions about: standardisation of each indicator, the
degree of substitution (value of [beta]), weights and inequality
aversion parameter (value of [delta]). Each outcome was divided with its
respective mean for standardisation, the value of [beta] was restricted
to zero because of ratio-scale invariance property. Cobb-Douglas
function was used and Value of parameter infers value judgment about
indicators and it is difficult to say which indicator is most important
in human welfare so equal weights are employed in analysis. (2)
Standard Gini coefficient possessed the value of [delta] equal to
2. This study computed Gini Coefficient for two values of delta one
value was equal to 2 and other was 5 which represent the higher weight
for bottom distribution. Higher the value of delta, higher weights will
be given to bottom distribution. Results are given below;
Table 2
Gini Coefficient in 2005-06 and 2010-11
Delta=2 Delta=5
Gini Coefficient 2005-06 2010-11 2005-06 2010-11
Expenditure .32 .29 .5 .45
Health Risk Index .19 .21 .34 .38
Max. Education .34 .33 .69 .68
Multidimensional .35 .34 .61 .61
Welfare Index .64 .65 .38 .39
Source: Author's own calculations.
Table 2 illustrates consumption inequality in different wellbeing
dimensions in Pakistan in 2005-06 and 2010-11. Value of Gini-coefficient
ranges from 0 to 1 where zero means perfect equality and 1 means perfect
inequality. Gini index shows the percentage of inequality prevailing in
a country. The table portrayed that consumption expenditure inequality
has declining trend from 0.32 to 0.29 for delta equal to two and more
severe when value of [delta] = 5 but inequality has decreased for both
values of delta. Health risk inequality has increased from 0.19 to 0.21
for delta equal to 2 and also increased from 0.34 to 0.38 for delta
equal to 5. Education inequality declined for delta equal to 2 and
remained same when value of delta is 5. Multidimensional inequality
showed slightly decline in dispersion when delta is 2 but showed no
significant reduction in dispersion for delta equal to 5. Welfare index
showed welfare has only 1 percent reduction for both values of delta and
welfare has declined about 26 percent with the increase in value of
delta from 2 to 5.
Upshot, results showed that multidimensional poverty also declined
from 51 percent to 36 percent. The dimensional breakdown of MP1 showed
that Immunisation, postnatal care, Enrolment and Equivalent per adult
expenditures comprise 66 percent of whole poverty in 2005-06, while in
2010-11 these four dimensions contribute 62 percent in overall poverty.
From 2005-06, Consumption inequality has declined about by 3 percent
while Multidimensional inequality has declined only by 1 percent.
3.4. Policy Recommendations
The results of present study give some suggestions about polices
for development. Policy for curbing poverty should be dimension
specific. Education and health contribute 62 percent to overall poverty.
Government should invest in health and education to provide more
facilities, to accelerate pace of poverty reduction, to redistribute and
improve health and educational facilities. Access to water has declined;
the proper mechanism must be taken for provision of water supply schemes
and improvement in the drainage system. The study opens some future
research avenues like to explore main causes of multidimensional poverty
and inequality. Additionally multidimensional poverty and inequality
should also be dissected at provincial and rural-urban levels.
REFERENCES
Ahmad, M. and N. S. Shirazi (2000) Estimation of Distribution of
Income in Pakistan, Using Micro Data. The Pakistan Development Review
39:4, 807-824.
Alkire, S. and J. Foster (2007) Counting and Multidimensional
Poverty Measures. (OPF1I Working Paper Series 7).
Alkire, S. and M. E. Santos (2010) Acute Multidimensional Poverty:
A New Index for Developing Countries. (OPHI Working Paper No. 11).
Aristei, D. and B. Bracalente (2011) Measuring Multidimensional
Inequality and Wellbeing: Methods and an Empirical Application to
Italian Regions. Statistica 71:2, 239266.
Atkinson, A. B. and F. Bourguignon (1982) Comparison of
Multidimensional Distribution of Economic Status. Review of Economic
Studies 49, 183-201.
Basu, K., and J. Foster (2008) On measuring Literacy. Economic
Journal 108, 1733-49.
Batana, Y. M. (2008) Multidimensional Measurement of Poverty in
Sub-Saharan Africa. (OPHI Working Paper No. 13).
Decancq, D. (2009) Essays on the Measurement of Multidimensional
Inequality. (Doctoral dissertation) Katholieke Universiteit Leuven,
Belgium.
Decancq, K. and M. A. Lugo (2012) Inequality of Wellbeing: A
Multidimensional Approach. Economica 79:316, 721-746.
Fisher, F. Mm (1956) Income Distribution, Value Judgements and
Welfare. Quarterly. Journal of Economics 70, 380- 424. v
Foster, J. E., J. Greer, and E. Thorbecke (1984) A Class of
Decomposable Poverty Indices. Econometrica 52:3, 761-766.
Gajdos, T. and J. A. Weymark (2005) Multidimensional Generalised
Gini Indices. Economic Theory 26:3, 471-496.
Jamal, H. (2009) Estimation of multidimensional poverty in
Pakistan. Social Policy and Development Centre. (Research Report No.
79).
Jorda, V., C. Trueba, and J. M. Sarabia (2013) Assessing Global
Inequality in Well-being Using Generalised Entropy Measures. Procedia
Economics and Finance 5,361-367.
Justino, P. (2005) Empirical Applications of Multidimensional
Inequality Analysis. University of Sussex, England. (Working Paper No.
16).
Kolm, S. (1977) Multidimensional Equalitarianisms. Quarterly
Journal of Economics 91, 1-13.
Lugo, M. A. (2004) On Multivariate Distributions of Well-being: The
Case of Argentine Provinces in the 1990s. Paper presented in General
Conference of IARIW, Cork, Ireland.
Mehta, K. A. and A. Shah (2003) Chronic Poverty in India:
Incidence, Causes and Policies. World Development 31:3, 491-511.
PBS (2006) Household Integrated Economic Survey 2005-06. Government
of Pakistan, Islamabad.
PBS (2011) Household Integrated Economic Survey 2010-11. Government
of Pakistan, Islamabad.
Rohde, N. and R. Guest (2013) Multidimensional Racial Inequality in
the United States. Social Indicators Research 114:2,591-605.
Sen, A. K. (1985) Commodities and Capabilities. North-Holland,
Amsterdam.
Sen, A. K. (1975) Minimal Conditions for Monotonicity of Capital
Value. Journal of Economic Theory 11:3, 340-355.
Sen, A. K. (1976) Poverty: An Ordinal Approach to Measurement.
Econometrica 44, 219-231.
Sen, A. K. (1985) A Sociological Approach to the Measurement of
Poverty: A Reply [Poor, Relatively Speaking]. Oxford Economic Papers
37:4, 669-76.
Seth, S. (2010) Essays in Multidimensional Measurement: Welfare,
Poverty, and Robustness. (Doctoral dissertation). Vanderbilt University,
USA.
UNDP (2005) Human Development Report; International Cooperation at
a Crossroads Aid, Trade and Security in an Unequal World; New York, USA.
Walzer, M. (1983) Spheres of Justice. New York.
World Bank (1990) Poverty World Development Report. Oxford
University Press.
Comments
* The paper has estimated the MPI and MGI.
* On page 1 author wrote, "Although measuring MPI is a
difficult task but due to the importance of MPI developed and developing
countries have adopted multidimensional poverty estimation". Though
2014 global MPI report measure more than 100 countries but it not
official, except few countries, Mexico, Colombia, Philippines, China,
Brazil, Bhutan, Malaysia, Indonesia, Chile
* Why study is measuring MPI and MGI, whether you have some policy
target and you mould MPI indicators as according to conventional MPI
methodology, you will suggest policy then. For example, in Colombia
country's development plan has three pillars: employment, poverty
reduction and security. The government plans to reduce multidimensional
poverty by 13 percent- from 35 percent of the entire population in 2008
to 22 percent in 2014. They put more emphasis on education and labour
indicators (long term employment, formal employment) to improve quality
of education and to resolve the issues of decent work. In Mexico, MPI is
the part of public policy in which government aim to pull the extreme
poor out of poverty based on both MPI and uni-dimensional approach.
* The study has utilised Pakistan Social and Living Standard
Measurement (PSLM) surveys (2005-06. and 2010-11). But you have reported
the sample of HIES data (15453 and 16341) so you not using PLSM data.
Though you can use if you not indicator from consumption and income
side.
* The MPI and MGI results are highly sensitive to the choice of
indicators. Author wrote 10 pages on theoretical modelling of MPI and
MGI while there is no discussion of selection of dimensions, indicators,
weights, cut-off which should be because all your results are based on
it that how you estimated the MPI and MGI.
* In result section 4.1.1. The author has stated to use four
dimensions; education, expenditure, living standard and health I have
reservation on expenditure dimension because it is usually determined by
the other 3 and our discussion with Alkaire in April, she suggest not to
use expenditure.
* But again there is no discussion on indicators before explaining
the results. Indicators should be representative, sustainable, and less
collinear to each other
* The results are very alarming in Figures 1 and 2, it should that
MPI is 51 percent in 2005 and only 12 percent in 2010. Two reservations.
Firstly we Pakistan not perform well health, education, expenditure and
living standard indicators. Second we have estimated MPI with Alkaire.
k-value 2004/05 2006/07 2008/09 2010/11
17 0284 0.272 0.261 0.230
23 0.264 0.252 0.240 0.208
28 0.240 0.229 0.217 0.185
33 0.219 0.208 0.197 0.166
34 0.211 0.200 0.188 0.157
40 0.179 0.169 0.158 0.129
50 0.122 0.117 0.108 0.085
52 0.105 0.102 0.094 0.072
62 0.057 0.057 0.050 0.037
75 0.016 0.017 0.014 0.010
* In Figures 3 and 4, author has mentioned the change in MPI within
the dimensions. I think there should also be explain the reasons.
* In Figure 5 author has mentioned the indicators. It should be
very earlier because whole your exercise is based on this analysis. But
still we don't know that what the definition of indicators is. Some
indicators look not representative, covering whole population (gas,
enrolment, or not sustainable.
* Finally your Table 1 and Table 2 is totally contradict to Figure
1 (51 percent) and Figure 2 (12 percent).
* You have too summarised the discussion on MGI. Your take 11 pages
to discuss MPI results while there is only 1 page on MGI. Again there is
methodological gap that how you calculated as you mention health risk
index, welfare index etc. but I don't know how you estimate these
variables.
* Finally there is no conclusion or policy recommendation.
Shujaat Farooq
Pakistan Institute of Development Economics, Islamabad.
(1) To attain universal primary education by 2015.
(2) HDI also used equal weights.
Maqbool H. Sial <maqsial@yahoo.com>, Asma Noreen, and Rehmat
Ullah Awan <awanbzu@ gmail.com> are affiliated with the Department
of Economics, University of Sargodha, Sargodha.
Table 1
Adjusted Headcount Ratio at Different Cut Offs in 2005-06 and 2010-11
2005-06
K H A [M.sub.o] = H x A
1 91.86 40.08 36.82
2 72.07 46.81 33.74
3 51.06 55.55 28.36
4 36.85 62.71 23.11
5 27.53 68.52 18.86
6 16.86 76.62 12.92
7 10.60 83.65 8.87
8 05.58 89.06 4.97
9 1.64 96.64 1.58
10 0 0 0
2010-11
K H A [M.sub.o] = H x A
1 60.02 36.55 25.23
2 54.40 42.07 22.89
3 35.86 50.56 18.13
4 22.60 58.66 13.26
5 15.37 64.53 9.92
6 8.60 73.09 6.29
7 3.84 80.71 3.10
8 .16 88.16 1.41
9 .10 96.46 .10
10 0 0 0
Source: Author's own calculations.
Fig. 2. Dimensional Share to National Poverty 2005-06 and 2010-11
2005-06 2010-11
Health 32 28
Education 23 28
Expenditure 20 16
Living Standard 25 18
Source: Author's own calculations.
Note: Table made from bar graph.
Fig. 3. Percentage of Household Deprived in each Indicator during
2005-06 and 2010-11
2005-06 2010-11
Yr. of Schooling 25.4 21.6
Enrollment 29 31.2
Postnatal Care 77 28
Immunization 25.5 19.3
Expenditures 22 2
Sanitation 45.3 42
Crowding 45.8 53.7
Gas 71.7 63
Electricity 14.3 9.4
Safe Drinking 15.3 16.2
Indicators
Source: Author's own calculations.
Note: Table made from bar graph.
