Multidimensional

Poverty in Pakistan

Acknowledgements

This report on Multidimensional Poverty has been developed in collaboration with the Oxford Poverty and Human Development Initiative (OPHI) and the United Nations Development Programme (UNDP), Pakistan.

The Ministry of Planning, Development & Reform is grateful to both OPHI and UNDP for their technical support in developing the report's methodology and analysis. We are also thankful to Pakistan Institute of Development Economics for facilitating provincial and regional consultations on the methodology and selection of indicators to construct the Multidimensional Poverty Index. We acknowledge the support and assistance of the Pakistan Bureau of Statistics in providing data and guidance. We also thank the Provincial and Regional Bureaus of Statistics for their participation in the consultation process and providing technical inputs.

This report is the result of wide-ranging consultations with different stakeholders from the public, private and other sectors. We would like to thank all the institutions and individuals who participated in these consultations and provided their feedback (a full list of names is included as an Annex).These discussions were instrumental in improving the methodology of the MPI and ensuring that it offers the best representation of poverty in the particular cultural, geographical and social context of Pakistan.

| i

Foreword

Preface

This report presents Pakistan's first official national Multidimensional Poverty Index (MPI). It marks the Government of Pakistan's endeavours to complement existing consumption-based poverty estimates with a non- income based approach to measuring poverty. It is intended that the MPI will provide evidence and a basis for public policy and resource allocations, especially under the National Finance Commission and the Provincial Finance Commission.

Pakistan's Vision 2025 prioritises investment in human capital and social services. It recognises the importance of inclusive and balanced growth - one which promotes the concept of shared prosperity and endeavours to address geographical and social inequality. The current Government strongly believes that the benefits of growth must be shared by all segments of society especially those from marginalized groups. The MPI will therefore serve as a useful instrument to guide public policies for inclusive growth and resource distribution.

This report provides evidence and analysis to align the Government's policies to the objective of reducing poverty in all its dimensions and addressing inequality. Vision 2025 stresses a broader definition of poverty – one which includes health, education and other amenities alongside income and consumption. It promises an increase in resource allocations to improve service delivery, governance and innovation in the economy. Consistent with these objectives, this report p r o v i d e s a d e t a i l e d a n a l y s i s o f t h e s i t u a t i o n o f multidimensional poverty in the country, as well as the different factors that have contributed to shaping it.

Over the past years, Pakistan's economy has grown. Today, an increasing proportion of population has access to healthcare services and education. A healthier economy will pave the way for improved employment opportunities and better standard of living. Women participation in social and economic spheres of life is increasing in Pakistan.

However, as demonstrated by the findings of this report, the economic gains have not translated into equal poverty reduction and prosperity across all regions and provinces of Pakistan. The resulting inequality has created a gap in development progress, with the depth and extent of poverty varying widely across the country. The MPI provides disaggregated statistics at the district level alongside in- depth information on the main contributors to poverty in all its dimensions. Thus, the MPI provides strong evidence for policy makers to identify the root causes of poverty and deprivation across Pakistan's regions and territories.

Furthermore, the analysis shows that some districts in Pakistan have lagged behind significantly in terms of social development, exhibiting high levels of poverty and deprivation. These districts should become priority areas for the Government to invest in social development and

| ii

accelerate the pace of overall development. This report also provides a trend analysis across different time periods. Such analyses are useful for assessing the impact of policies and for identifying gaps.

The report is timely in the wake of adoption of Sustainable Development Goals by Pakistan, the local government elections and devolution resulting from Pakistan's 18th Constitutional Amendment. The district level analysis of Pakistan's MPI will aid local governments in identifying sectors that require greater attention, enabling them to allocate resources accordingly. It will also provide useful analysis to identify and address development challenges at micro and macro level. I hope this report will generate dialogue and further research to deepen our understanding of the key drivers of poverty in Pakistan.

T h e P l a n n i n g C o m m i s s i o n w i l l u s e t h e M P I a s a complementary measure of poverty along with the consumption based poverty measure and will encourage provinces and local governments to use it for their policy interventions for poverty targeting and inequality reduction. I trust that this report will also be used by all relevant stakeholders as a tool to design their interventions and track progress.

I would like to express my particular appreciation for Dr Naeem uz Zafar, Member, Social Sector and Mr Zafar ul Hassan, Chief Pover ty and SDGs Section, Planning Commission for leading the preparation of this report. I acknowledge the participation of Provinces and Regions in the consultation process and commend their invaluable feedback. I thank Mr Shakeel Ahmad, Assistant Country Director, UNDP and his team and Professor Sabina Alkire, Director, Oxford Poverty and Human Development Initiative and her team for their technical support. I also highly appreciate the continuous support of the United Nations Development Programme (UNDP) to the Ministry of Planning, Development and Reform.

Professor Ahsan Iqbal Federal Minister

Ministry of Planning, Development & Reform, Pakistan

Poverty is a complex and multidimensional phenomenon. It is often said that poverty is an elusive concept and it is hard to decide that poverty is output of some endowments and choices or it is input to metrics of better well-being. This duality helps in understanding the basic difference between money metric poverty, which is primarily an outcome based measure, and Multidimensional Poverty, which is primarily an input based measure. Multidimensional poverty is based on several deprivation such as the inability to attain a good education, a lack of access to healthcare facilities, poor housing and an unsafe environment in which to live. The index computed by aggregating these deprivations has profound usefulness for policies and plans as this index can be disaggregated on basis of deprivations and geography. This suggests that Multidimensional poverty is helpful for balanced social policies.

The Global Multidimensional Poverty Index (MPI), originally established by the Oxford Poverty and Human Development Initiative (OPHI), University of Oxford, and the United Nations Development Programme (UNDP), is a measure that integrates the wider concept of poverty by reflecting on deprivations experienced by individuals with respect to health, education and standard of living. Therefore, it serves as a useful tool for public policy. Since the inception of this index in 2010, many countr ies have adapted the methodology behind the Global MP and created an official multidimensional poverty estimate, usually complementing consumption- or income-based poverty figures. The use of the MPI is as relevant to the context of Pakistan as it is to other countries.

This repor t marks the first time that estimates of multidimensional poverty in Pakistan have been provided at the national, provincial and district levels. It also includes a trend analysis spanning 2004-2015. The reduction of multidimensional poverty is one of the core objectives of Pakistan's Vision 2025. This report thus establishes a baseline not for only Vision 2025, but also for Pakistan's progress towards the Sustainable Development Goals. The report provides a retrospective understanding of Pakistan's progress over more than a decade. As the report compares poverty across provinces, regions and districts, Pakistan's official MPI constitutes a useful tool for targeting as well as for detecting and addressing spatial inequalities and other group-based disparities.

Led by the Ministry of Planning, Development & Reform, this report is the product of wide ranging consultations involving Pakistan's Federal and Provincial Government Ministries and D e p a r t m e n t s , a c a d e m i a , r e s e a r c h o r g a n i s a t i o n s , development partners and other stakeholders. Technical inputs for the report were provided by OPHI and UNDP.

I appreciate the contributions of UNDP and OPHI in terms of their technical assistance and support in compiling the findings of this report. I also gratefully acknowledge the input of academia and the useful feedback of the provinces which participated in consultations to inform the report.

Dr Naeem uz Zafar Member, Social Sector

Planning Commission of Pakistan

| iii

Message from UNDP Pakistan

Message from OPHI,

University of Oxford

It is with immense pleasure that we celebrate the launch of Pakistan's first ever National Report on Multidimensional Poverty by the Ministry of Planning, Development and Reform.

This report is particularly timely in the first year of implementation of the Sustainable Development Goals (SDGs). Multidimensional poverty estimates can help establish solid baselines for tracking progress towards these new global goals for poverty alleviation and sustainable development, and particularly on SDG10, 'to reduce inequality within and among countries'.

Poverty has declined globally though mostly driven by China. The complete elimination of poverty by 2030 is considered to be within reach. However, inequality within and between countries has increased and is considered to be the key development challenge of the 21st century. Similarly, in Pakistan, poverty has declined but inequality has worsened. Because of its importance, “leaving no one behind” is one of the key objectives of the SDGs. In this context, the multidimensional poverty estimates especially at the sub-national level will be extremely helpful in identifying deprived geographical areas and communities and informing public policy for improved targeting.

Following the 18th Constitutional Amendment, Pakistan's governance structure has been largely devolved to the provinces, which now take the lead in many development interventions and are supported by an emergent local government structure. In this context, the report provides disaggregated data at the district level which will be invaluable for local authorities in tracking deprivation, and targeting poverty eradication measures and achieving the SDGs in their respective districts.

The Multidimensional Poverty is intended to serve as a complementary measure to consumption / income based poverty estimates. As it measures deprivations experienced by individuals in health, education and standard of living, it complements the consumption / income based poverty by reflecting upon other non-monetary facets of poverty. Together, the consumption based poverty estimates and multidimensional poverty provide an insightful and detailed picture of the different forms of monetary and non-monetary deprivation that people are suffering from.

M a n y c o u n t r i e s a c r o s s t h e g l o b e a r e u t i l i z i n g multidimensional poverty as a tool for planning, budgeting and targeting the marginalized segments of society. In Pakistan's context, it could be used for informing allocations to the most deprived regions of Pakistan under the National and Provincial Finance Commission awards. It can also inform government's policies on social protection and gender equality.

In light of the importance and utility of the multidimensional poverty index as a tool for public policy, we at UNDP are pleased to par tner with the Ministr y of Planning, Development and Reform, alongside the Oxford Poverty and Human Development Initiative at the University of Oxford, in preparing this report. We are committed to providing similar support in future and continuing this important partnership towards the achievement of Pakistan's SDGs.

Marc-André Franche Country Director

United Nations Development Programme

In 'Antesaab', by Faiz Ahmed Faiz, translated by Mahbub ul Haq, we are reminded of the following dedication:

To

This day

And the deep pain of this day: A pain that is a silent insult

To the false glamour of life around…

Pakistan's MPI is, in many ways, quietly seeking to advance such a dedication in the present day, under the leadership of the Planning Commission, and in partnership with UNDP. In order to benefit from the wisdom of many actors, Planning Commission and UNDP staff convened leaders in government, academia, civil society, and other sectors through provincial level consultations to think about no other topic than, 'the deep pain of this day', and to articulate in a constructive and empowering manner.

Built using the PSLM datasets, the MPI has been estimated for e ve r y t wo - ye a r p e r i o d s i n ce 2 0 0 4 / 5 , a n d c a n b e disaggregated by both provincial and district levels. This feature enables Pakistan's MPI to be used as a tool for planning and management – because it is updated often enough to see change, and because it provides information to lower levels of government as well as to national institutions.

Because Pakistan's MPI can be unfolded to see how people are poor – the deprivations they experience in a given district, province, or social group – it can also be a tool of policy coordination, and of budget allocation. And because Pakistan's MPI was assessed using a series of robustness tests (Annex 2), which found the analysis based on Pakistan's MPI to be robust to a plausible range of weights and poverty cut- offs, it can be commended as a suitably rigorous measure for policy purposes.

Pakistan's MPI can be used to diagnose the places in which poverty is the highest, and to show how people are poor in different areas. This information might be useful to non governmental organisations and civil society groups who are interested to fight poverty in their focal areas, or private s e c to r a c to r s w h o a re p l a n n i n g co r p o r a te s o c i a l responsibility activities or philanthropic investments.

Pakistan's MPI design also contains some hidden gems. For example, because of a commitment to gender equity in education, it is not enough only to have an educated man. Pakistan's MPI views a household as not having achieved sufficient years of schooling unless at least one woman and one man above 10 years of age has completed 5 years of schooling. Similarly, Pakistan's MPI prioritises women's ante- natal care and safe deliveries, and considers quality

education of both girls and boys to be paramount. So insofar as the historical data permit, the MPI integrated women's agency within its very design.

Pakistan is a member of the Multidimensional Poverty Peer Network (MPPN.org), a South-South network of over 40 countries plus international agencies. Many countries in the network are using national MPIs to energise their fight against poverty in all its dimensions, and to renew their solidarity with the disadvantaged. Our hope is that Pakistan's MPI will fuel not controversy but compassion. That it will burst apathy and kindle commitment. And by using the MPI to fight human disadvantage with innovation and determination, Pakistan will chart a path that other nations too, will wish to follow.

Professor Sabina Alkire Director, Oxford Poverty & Human Development Initiative

University of Oxford

| iv | v

Contents

Acknowledgements Foreword

Preface

Message from UNDP Pakistan

Message from OPHI List ofTables

List of Figures Acronyms Executive Summary

1

Chapter 1

Introduction

Money Metric Poverty Measure in Pakistan

Context and Framework Purpose of the MPI Measure

i ii

iii iv

v viii ix

x xi

7

Chapter 2

Methodology

Measurement Design

Unit of Identification and Analysis Dimensions, Indicators and Cut-offs Weights

Alkire Foster Methodology

The Multidimensional Poverty Index: an Adjusted Headcount Ratio

Properties of the Multidimensional Poverty Index

Poverty and Deprivation Cut-offs

Data

13

Chapter 3

Main Results

National Uncensored Headcount Ratios

Pakistan's National MPI – Key Results The Composition of Poverty:

Percentage Contributions of Each

Indicator to the MPI

23

Chapter 4

Changes in Multidimensional Poverty Over Time

Changes in National Uncensored Headcount Ratios

Changes in Multidimensional Poverty Index and its Components Over Time

Changes in National Censored Headcount Ratios

35

Chapter 5

Multidimensional Poverty at District Level

47

Chapter 6

Conclusion

53

Statistical Annex

Annex 1: Reader's Guide to the Alkire-Foster Methodology

Annex 2: Robustness Analysis Annex 3: Statistical Tables

References 112

List of Tables

Table 1.1: Official Poverty Rates in Pakistan, 1998/99 – 2013/14

Table 2.1: Pakistan's National MPI – Indicators, Deprivations, Cut-offs and Weights Table 3.1: Incidence, Intensity and Multidimensional Poverty Index (MPI), 2014/15 Table 3.2: Multidimensional Poverty by Rural/Urban Areas, 2014/15

Table 3.3: Multidimensional Poverty by Province/Region

Table 3.4: Confidence Interval for Provincial/Regional Multidimensional Poverty

Table 3.5: Percentage Contributions of Indicators to MPI at the National and Provincial/Regional Level

Table 4.1: Change overtime in Incidence, Intensity and the MPI, 2004-2015

Table 4.2: Statistical Significance of Change in Headcount for All Provinces

List of Figures

Figure 3.1: National Uncensored Headcount Ratios Figure 3.2: Multidimensional Poverty Index (MPI) Figure 3.3: Headcount (H)

Figure 3.4: Intensity (A)

Figure 3.5: Percentage Contribution of Each Indicator to MPI, by National and Rural/Urban

Figure 3.6: Percentage Contribution of Each Indicator to MPI, by Province Figure 3.7: Percentage Contribution of Each Indicator to MPI, by Region Figure 4.1: National Uncensored Headcount Ratios, 2004-2015 Figure 4.2: National MPI, 2004-2015

Figure 4.3: National Incidence (H), 2004-2015 Figure 4.4: National Intensity (A), 2004-2015 Figure 4.5: Punjab MPI, 2004-2015 Figure 4.6: Punjab Incidence (H), 2004-2015 Figure 4.7: Punjab Intensity (A), 2004-2015 Figure 4.8: Sindh MPI, 2004-2015

Figure 4.9: Sindh Incidence (H), 2004-2015 Figure 4.10: Sindh Intensity (A), 2004-2015 Figure 4.11: Khyber Pakhtunkhwa MPI, 2004-2015

Figure 4.12: Khyber Pakhtunkhwa Incidence (H), 2004-2015 Figure 4.13: Khyber Pakhtunkhwa Intensity (A), 2004-2015 Figure 4.14: Balochistan MPI, 2004-2015

Figure 4.15: Balochistan Incidence (H), 2004-2015 Figure 4.16: Balochistan Intensity (A), 2004-2015 Figure 4.17: Absolute Change in MPI, 2004-2015 Figure 4.18: Relative Change in MPI, 2004-2015 Figure 4.19: Rural Areas' MPI, 2004-2015 Figure 4.20: Rural Areas' Incidence (H), 2004-2015 Figure 4.21: Rural Areas' Intensity (A), 2004-2015 Figure 4.22: Urban Areas' MPI, 2004-2015 Figure 4.23: Urban Areas' Incidence (H), 2004-2015 Figure 4.24: Urban Areas' Intensity (A), 2004-2015

Figure 4.25: National Censored Headcount Ratios, 2004-2015

Figure 4.26: Change in Censored Headcount Ratios, 2004-2015

Figure 5.1: Starting MPI Value vs Absolute Reduction of MPI by District, 2004-2015

Figure 5.2: Absolute Change in Headcount, 2004-2015

Figure 5.3: Relative Change in Headcount, 2004-2015

| viii

| ix

Acronyms

Executive Summary

AJK Azad Jammu & Kashmir BHU Basic Health Unit

CBN Cost of Basic Needs

FATA Federally AdministeredTribal Areas FEI Food Energy Intake

GB Gilgit-Baltistan

HDI Human Development Index HDR Human Development Report KP Khyber Pakhtunkhwa

MPI Multidimensional Poverty Index

OPHI Oxford Poverty and Human Development Initiative, University of Oxford PSLM Pakistan Social and Living Standards Measurement

UNDP United Nations Development Programme

| x

Pakistan's Vision 2025 reaffirms the need to make economic growth inclusive and sustainable in order to eradicate poverty. It also recognises that poverty is multidimensional, encompassing not only monetary deprivation but also the inaccessibility of healthcare, education and other amenities for all communities across the country.

In accordance with the Government's commitment to eradicate poverty, this report presents Pakistan's first national Multidimensional Poverty Index (MPI) based on the Alkire- Foster methodology. It has three dimensions: education, health and living standards. To tailor the measure to Pakistan's context and public policy priorities, 15 indicators were used for this national measure, instead of the 10 employed for the global measure. Within these 15 indicators, three indicators are included under the dimension of education (years of schooling, child school attendance, and educational quality), four under health (access to health facilities/clinics/Basic Health Units (BHU), immunisation, ante-natal care, and assisted delivery) and eight under living standards (water, sanitation, walls, overcrowding, electricity, cooking fuel, assets, and a land/livestock indicator specifically for rural areas). Each of the three dimensions carries an equal weight of 1/3 of the MPI. The weights of the component indicators within each dimension are equal unless another justification is provided, as outlined in Section 2.1.3. Overall, a person must be deprived in 1/3 of these weighted indicators to be identified as multidimensionally poor.

Multidimensional Poverty at a Glance

Applying this measure to data from the Pakistan Social and Living Standards Measurement (PSLM) survey for the 2 0 1 4 / 1 5 p e r i o d , w e f o u n d t h a t t h e c o u n t r y ' s Multidimensional Poverty Index stands at 0.197. This indicates that poor people in Pakistan experience 19.7% of the deprivations that would be experienced if all people were deprived in all indicators. Secondly, it must be noted that the MPI is a product of two essential components: the poverty “headcount” and the “intensity” of deprivation. Using the same data from the 2014/15 PSLM survey, the country's multidimensional poverty“headcount ratio”was estimated at 38.8% of the population. This means that 38.8% of the population of Pakistan are poor according to the MPI. The average intensity of deprivation, which reflects the share of deprivation which each poor person experiences on average, is 50.9%.

There are stark regional disparities in poverty across Pakistan. The proportion of people identified as multidimensionally poor in urban areas is significantly lower than in rural areas – 9.4% and 54.6%, respectively. Further heterogeneities were found when looking at results at the provincial level. In 2014/15, MPI headcount ratios ranged from 31.4% in Punjab (with an intensity of 48.4%), to 71.2% in Balochistan (with an average intensity of 55.3%).

With respect to the percentage which each of the 15 indicators contributes to overall multidimensional poverty in

Pakistan, the greatest contribution to national poverty derives from years of schooling (29.7%), followed by a lack of access to healthcare facilities (19.8%) and child school attendance (10.5%). If aggregated by dimensions, the greatest contribution to poverty stems from educational deprivation (42.8%), followed by living standards (31.5%) and healthcare (25.7%).

Reductions in Multidimensional Poverty OverTime

Since 2004/05, multidimensional poverty has continuously declined in Pakistan. The MPI fell from 0.292 in 2004/05 to

1. in 2014/15, while the poverty headcount ratio fell from 55.2% to 38.8%. The intensity of deprivation also declined over the same period, falling from 52.9% to 50.9%. Similar trends are evident across all provinces and regions, with the exception of A zad Jammu & K ashmir (A JK ) which experienced an increase in multidimensional poverty between 2010/11 and 2012/13. In terms of relative change in its MPI, Punjab accounts for the highest relative reduction (40.2%), while Balochistan experienced the slowest progress in reducing multidimensional poverty, with a relative change of only 17.7%.

   
   

   At the district level, Larkana, Attock, Malakand, T.T. Singh and Hyderabad have made the most progress, reducing absolute poverty headcount ratio by more than 32 percentage points. In relative terms the best performers were the districts of Islamabad, Attock, Jhelum, Lahore, Karachi and Rawalpindi. On the other hand, some districts have experienced an increase in their poverty incidence. In absolute and relative terms, the districts of Umerkot, Harnai, Panjgur, Killa Abdullah and Kashmore have witnessed the highest increase in incidence of poverty.

   
   

   This report provides a detailed description of these results and disaggregates Pakistan's MPI by indicators, geographical regions and sub-groups. While the report closes with a series of specific recommendations, all of the findings are provided with the intention to help the Federal and Provincial Governments in targeting poverty through improved policy reform and public spending.

   
   

   | xi

   

   

   
   

   Chapter 1 Introduction

   
   

   Money Metric Poverty Measure in Pakistan

   Context and Framework Purpose of the MPI Measure

   

   Chapter 1 Introduction

   
   

   A measure of multidimensional poverty is a natural progression given Pakistan's history of economic development and its trajectory of social indicators. Between 1990 and 2013, Pakistan's GDP per capita in constant 2005 US Dollars increased from USD 542 to 793, with growth rates averaging around 4% per year.1 Until 2003, despite periods of instability, Pakistan was ahead of both India and Bangladesh in terms of its GDP per capita. Moreover, income-based poverty fell sharply in the country, with the percentage of the population living below the national poverty line decreasing from 64.3% in 2001/02 to 29.5% in 2013/14.2 In fact, by 2005 Pakistan had already met its Millennium Development Goal of halving the percentage of people who were “income poor” with respect to the USD 1.25/day poverty line.

   
   

   However, similar progress has not been evident across vital social indicators. According to World Bank’s World Development Indicators, despite rapid improvements in immunisation Pakistan still lags behind coverage rates in South Asia. Compared to Bangladesh, Pakistan started out much better in terms of life expectancy (60 years in 1990) and was second only to Sri Lanka in this respect. Yet, by 2014 life expectancy in Pakistan had merely increased to 66 years. By contrast, the improvement in Bangladesh was far greater, with life expectancy rising from 58 to 72 years during the same period. Similarly, Pakistan's infant mortality rate (IMR) was slightly above that of Bangladesh in 1990, at 106 deaths per 1,000 (as opposed to 100 in Bangladesh). Unfortunately, by 2015 Pakistan was still registering the deaths of 66 infants in their first year, as opposed to 31 in Bangladesh. In fact, Pakistan along with Afghanistan currently have the highest IMR rates of any country in South Asia, all of which register fewer than 50 infant deaths per 1,000. Comparable patterns hold true for maternal mortality, as Pakistan began ahead of all other South Asian nations – with the exception of Sri Lanka – but now has higher rates than most of the other countries in the region. Furthermore, fertility rates in Pakistan were – and remain, one of the highest in South Asia at 3.6 children per woman.3

   
   

   This situation has been well-noted by many actors within Pak istan. Introducing Pak istan's Vision 2025 National Development Plan, President Mamnoon Hussein pointed out that the Plan:

   highlights the imbalance between economic development

   trends: “Today, we find many countries which were lagging behind have forged ahead and overtaken us.”4

   
   

   To re-balance Pakistan's portfolio of achievements, Vision 2025 specifically sets out to invest in lagging social sectors:

   While economic indicators situate the country among lower middle-income economies, the social indicators are comparable to those of least developed countries. The result is a fractured socio-economic platform for development. In order to become a developed nation, it will be necessary to redress this imbalance by giving top priority to building a strong human and social capital base as a prerequisite for sustainable development.

   
   

   The Plan's first Pillar, “People First: Developing social and human capital”, identifies strengthening human capital as “the foremost priority of Vision 2025.” It continues, “Recognizing the size and scale of this endeavour, we conceive a very significant increase in resource allocation, and quantum improvement in the quality of service delivery through good governance and innovation.”

   
   

   Vision 2025, in a manner consistent with these priorities, also broadens the definition of poverty to include health, education and other amenities alongside income or consumption:

   Pakistan Vision 2025 is people centric and aimed at reducing poverty and enhancing the people's well-being. Poverty is a multidimensional phenomenon and is described as a lack of income or consumption and access to education, health and other amenities of life.

   
   

   Pakistan's Multidimensional Poverty Index (MPI) has been developed as a tool to enable development actors in the country make significant progress on social indicators, reduce multidimensional poverty, and advance Pillar I of Vision 2015, as well as other social goals. Evidently, Pakistan's MPI clearly reflects Vision 2025. At the same time, its structure has been vetted and improved by groups of citizens, experts and leaders across all provinces. As such, it also seeks to enable the private sector, philanthropic and NGO actors to“crowd in”and play their part.5

   
   

   Pakistan's MPI can serve as a tool for good governance – for policy coordination, monitoring and readjusting programming, and for targeting and designing integrated policies that

   and social development, and suggests policies for

   accelerate progress.6

   The effectiveness of such policies is

   improving the socioeconomic indicators of the country. The turnaround from the current state of affairs in most social development indicators – including population welfare, poverty, gender mainstreaming, literacy, school enrolment, immunisation coverage and access to potable water – is promised by investing more in human and social development.

   
   

   The Minister of Planning and Lead Author of the Plan, Professor Ahsan Iqbal, also candidly acknowledged the aforementioned

   stressed in the preamble to the Sustainable Development Goals. Entitled Transforming Our World, the document highlights that “the interlinkages and integrated nature of the Sustainable Development Goals are of crucial importance in ensuring that the purpose of the new Agenda is realised”. This builds upon the UN Secretary-General's evidence-based hope that the SDGs will “inject new impetus for embracing integrated approaches to development.”. 7

   
   

   1Data from the World Bank's World Development Indicators.

   1. Pakistan Economic Survey, 2015/16.

   2. Data for all social indicators for South Asian countries have been taken from World Bank's World Development Indicators.

   3. Vision 2025, page ix.

5Costa Rica and Colombia are among the countries with a strong private sector contribution to reducing their national MPI.

6Pakistan is a member of the 40-country Multidimensional Poverty Peer Network (www.mppn.org) which includes many examples of states using their national MPI to manage and accelerate change, such as Mexico, Colombia and the Philippines.

7 A/70/75-E2015/55. Available: http://www.un.org/ga/search/view_doc.asp?symbol=A/70/75&Lang=E

Chapter 1 Introduction | 3

1.1 Money Metric Poverty Measure in Pakistan

Pakistan's official consumption-based poverty measure is currently under the scope of the Ministry of Planning, Development & Reform. The Ministry measures poverty using data from the Household Integrated Economic Survey, combined with the Pakistan Social and Living Standards Measurement survey (HIES/PSLM).

The estimates of poverty are produced by Planning Commission using the 'cost of basic needs' (CBN) methodology. Until recently, the approach used to estimate the headcount poverty in Pakistan was based on the food energy intake (FEI) methodology. Using the new CBN methodology, however, the poverty line has been revised from PKR 2,259.44 to PKR 3,030.32, per adult per month. Although this newly established poverty line is marginally higher, data still corroborates the decline in poverty trends in Pakistan. However, now a higher proportion of the population (29.5%) is considered to be below poverty line. The methodological revisions in monetary poverty reflect the government's expanded commitment to use improved measurement tools to identify and address poverty.

As Table 1.1 illustrates, the official monetary poverty rates in Pakistan experienced a strong decline between 1998/99 and 2013/14. In particular, the proportion of people living below the official poverty line dropped from 57.9% to 29.5% (a relative reduction of almost 49%). This marked decline may be associated with a number of factors, including increased allocations to social safety net programmes such as Benazir Income Support Programme (BISP). It may also be tied to better support prices for agricultural products, an improvement in the inflow of remittances, and increases in female labour force participation rates in rural areas.8

In addition, Table 1.1 identifies sizable disparities between rural and urban areas during this time period. Although both areas experienced a stark reduction in their poverty rates, rural areas still experience much higher levels of poverty than urban centres. Moreover, while poverty was 1.4 times higher in rural areas in 1998/99 than it was in urban areas (63.4% and 44.5%, respectively), this ratio increased to 1.95 in 2013/14 (with poverty rates of 35.6% and 18.2% for rural and urban areas, respectively).

Analysing poverty through monetary based measures alone suggests significant improvements in the country over the past decade. However, these have not resulted in an equal reduction

Table 1.1

Official Poverty Rates in Pakistan, 1998/99 – 2013/14

(% of the population living below the national poverty line)

Year National Urban Rural

Source: Planning Commission estimates using HIES/PSLM data (Ministry of Planning, Development & Reform, 2016)

of multidimensional poverty across the board. Thus, this report intends to use the multidimensional poverty analysis to complement the monetary poverty analysis, and reveal the true state of poverty in the country, bearing in mind the country's particular geographical and cultural context.

1.2 Context and Framework

Until recently, many countries measured poverty solely by taking i n t o a c c o u n t i n c o m e o r c o n s u m p t i o n . H o w e v e r, a unidimensional indicator like income cannot capture the multiple aspects of poverty. The global Multidimensional Poverty Index (MPI) is a new international measure of acute poverty developed by OPHI and UNDP's Human Development Report Office (UNDP HDRO). The MPI complements global monetary poverty measures by reflecting the acute deprivations that individuals simultaneously face in other dimensions. Like monetar y considerations, these are also essential to g u a ra nte e i n g a d i gn i f i e d l i fe. Fo l l ow i n g t h e H u m a n Development Index (HDI), the MPI shares the same three core dimensions: education, health and living standards. However, it expands on the number of indicators employed.

The MPI is based on the concept of capability, which is central to the human development paradigm championed by Mahbub ul Haq. Nobel Laureate, Professor Amartya Sen has argued that social evaluation should be based on the extent of people's freedoms to further the objectives that they value. The term “capability” or “capability set” provides information on the range of functioning that a person may reasonably achieve. Poverty in this framework becomes a “capability failure” – i.e. people's lack of capability to enjoy the key“beings and doings”that are basic to human life.This concept is inherently multidimensional.

The first global MPI was released in 2010 and measures acute poverty using a structure that can be compared across 75% of the global population, by country and population group. The global MPI has been updated regularly and published in every subsequent Human Development Report. Furthermore, OPHI's website (www.ophi.org.uk) features detailed tables, graphics, policy briefs and academic papers on the Index. However the global MPI was, from the start, developed with the secondary aim of encouraging the development of national versions of the MPI, tailored to specific national circumstances. Therefore, just as most countries have national income poverty measures which are used to inform policy (although the $1.90/day measure is used to compare countries), the aim is for interested countries to develop national MPIs that reflect their own development plans, data sources and aspirations.

1.3 Purpose of the MPI Measure

The analysis contained in this report represents an attempt to construct a national baseline for Pakistan's MPI that can be used as a yardstick against which to measure the progress of development in the coming years. In order to provide comprehensive, in-depth analysis, multidimensional poverty at the national level is disaggregated at the provincial/regional and district levels across different time periods.

The micro-level poverty data presented here may be used by Pakistan's Federal and Provincial Governments as a tool to target spatial inequalities and eliminate poverty in all its dimensions. It can help Governments assess how their policies are affecting people, particularly the poor. Given its availability at the provincial and district levels, the MPI can inform the poverty criteria of the National Finance Commission Award, as well as the

criteria of the Provincial Commission Awards. Newly established local governments can also use the MPI to inform their development interventions. The Planning Commission intends to produce estimates of multidimensional poverty either annually or once every two years, at both the national and sub- national levels. These estimates will not only be useful for development planning, but will also be used to track Pakistan's progress towards the Sustainable Development Goals, especially SDG 1, Target 1.2, which concerns the reduction of poverty in all its dimensions.

8 Ministry of Planning, Development & Reform, 2014.

4 Multidimensional Poverty in Pakistan

Chapter 1 Introduction | 5

Chapter 2 Methodology

Measurement Design

Unit of Identification and Analysis Dimensions, Indicators and Cut-offs Weights

Alkire Foster Methodology

The Multidimensional Poverty Index: an Adjusted Headcount Ratio

Properties of the Multidimensional Poverty Index

Poverty and Deprivation Cut-offs

Data

Chapter 2 Methodology

The methodology used in this report to determine Pakistan's MPI is adopted from Alkire and Santos' (2010, 2014) work on the global MPI, undertaken in collaboration with UNDP. This chapter outlines the report's methodology, describes the MPI and its relevant properties, and presents the data used to derive Pakistan's MPI. It comprises the following sections:

2.1 Measurement Design

Pakistan's national MPI utilises a set of dimensions, indicators and cut-offs that reflect its priorities as expressed in the Government's National Plans, and which can be implemented using the PSLM survey dataset. This section elaborates on the choice of these parameters.

2.1.1 Unit of Identification and Analysis

The unit of identification refers to the entity identified as poor or non-poor – usually the individual or the household. In the case of Pakistan's MPI, the unit of identification is the household. Information on the members of a household is considered together, all of whom receive the same deprivation score. This acknowledges intra-household caring and sharing. For example, educated household members reading to others, or multiple members being affected by the severe health conditions of a single member of their household. As such, this allows the measure to include indicators that are specific to certain age groups or genders, for instance, school attendance or ante-natal care.

The unit of analysis in which results are reported and analysed, however, is the individual. This means that, for example, the headcount ratio denotes the percentage of people who are

Table 2.1

Pakistan's National MPI – Indicators, Deprivation Cut-offs and Weights

identified as poor, rather than the percentage of households identified as poor, thereby valuing each citizen equally.

2.1.2 Dimensions, Indicators and Cut-offs

Pakistan's MPI builds upon the global MPI, retaining the same three core dimensions: education, health and living standards. The choice of indicators, however, reflects the country's particular context and political priorities, as well as the data available in the PSLM surveys. In total, 15 indicators are used in this national index, of which 7 indicators are the same as those used in the global MPI.

While the global MPI's health dimension includes the indicators of child mortality and nutrition, Pakistan's MPI does not have these indicators as they are not covered by PSLM survey. Instead, it uses the indicators of access to health facilities, full immunisation, ante-natal care, and assisted delivery. A noteworthy feature of Pakistan's 'years of schooling' indicator within the education dimension is the use of an innovative gendered component. This requires that at least one man and one woman in the household above the age of 10 has completed a minimum of 5 years of schooling. Finally, the national MPI also adds indicators concerning improved walls (instead of floors), overcrowding, and land/livestock to the living standards dimension. Details of the dimensions and indicators used in Pakistan's MPI are presented inTable 2.1.

Some of the indicators in Pakistan's MPI are clearly designed to support gendered understandings of poverty, such as ante-natal care and attended delivery. However it is important to note that the school attendance variable supports gender equity since if a boy or a girl is out of school the household is deprived. This is even more an emphasis in the years of schooling variable. In this

Dimension

Indicator Deprivation Cut-off Weights

Education

Health

Years of schooling Child school attendance School quality

Access to health facilities/

Deprived if no man OR no woman in the household above 10 years of age has completed 5 years of schooling Deprived if any school-aged child is not attending school (between 6 and 11 years of age)

Deprived if any child is not going to school because of quality issues (not enough teachers, schools are far away, too costly, no male/female teacher, substandard schools), or is attending school but remains dissatisfied with service

Deprived if health facilities are not used at all, or are only used once in a while, because of access constraints (too far away, too costly, unsuitable, lack of

1/6 = 16.67%

1/8 = 12.5%

1/24 = 4.17%

1/6 = 16.67%

clinics/ Basic Health Units (BHU) tools/staff, not enough facilities)

Standard of Living

Immunisation Ante-natal care

Assisted delivery Water

Sanitation Walls Overcrowding

Electricity Cooking fuel Assets

Land and livestock (only for rural areas)

Deprived if any child under the age of 5 is not fully immunised according to the vaccinations calendar (households with no children under 5 are considered non- deprived)

Deprived if any woman in the household who has given birth in the last 3 years did not receive ante-natal check-ups (households with no woman who has given birth are considered non-deprived)

Deprived if any woman in the household has given birth in the last 3 years attended by untrained personnel (family member, friend, traditional birth attendant, etc.) or in an inappropriate facility (home, other) (households with no woman who has given birth are considered non-deprived)

Deprived if the household has no access to an improved source of water according to MDG standards, considering distance (less than a 30 minutes return trip): tap water, hand pump, motor pump, protected well, mineral water

Deprived if the household has no access to adequate sanitation according to MDG standards: flush system (sewerage, septic tank and drain), privy seat Deprived if the household has unimproved walls (mud, uncooked/mud bricks, wood/bamboo, other) Deprived if the household is overcrowded (4 or more people per room)

Deprived if the household has no access to electricity

Deprived if the household uses solid cooking fuels for cooking (wood, dung cakes, crop residue, coal/charcoal, other) Deprived if the household does not have more than two small assets (radio, TV, iron, fan, sewing machine, video cassette player, chair, watch, air cooler, bicycle)

OR no large asset (refrigerator, air conditioner, tractor, computer, motorcycle), AND has no car.

Deprived if the household is deprived in land AND deprived in livestock, i.e.:

1. Deprived in land: the household has less than 2.25 acres of non-irrigated land AND less than 1.125 acres of irrigated land

2. Deprived in livestock: the household has less than 2 cattle, fewer than 3 sheep/goats, fewer than 5 chickens AND no animal for transportation (urban

households are considered non-deprived)

1/18 = 5.56%

1/18 = 5.56%

1/18 = 5.56%

1/21 = 4.76%

1/21 = 4.76%

1/42 = 2.38%

1/42 = 2.38%

1/21 = 4.76%

1/21 = 4.76%

1/21 = 4.76%

1/21 = 4.76%

Chapter 2 Methodology | 9

case, a household is deprived unless one woman and one man above 10 years of age have completed 5 years of schooling. This variable captures the gendered disadvantages in education. For example, 18% of people live in a household where no man or woman has completed five years of schooling. But where one gender has and the other has not, the difference is clear: only 4.8% of people live in a household where a woman but no man has completed five years of schooling, whereas 25.6% (more than five times as much) of the population live in a household where a man has completed five years of schooling, but no woman has had this opportunity. Reducing this deprivation – which contributes most to Pakistan's MPI – requires an investment in women's education, perhaps including life-long learning opportunities.

The selection of the dimensions, indicators, deprivation cut-offs and weights of Pakistan's MPI was based on thorough discussions and provincial consultations with government officials, academics, civil society organisations and experts in the field.

These decisions were later checked against existing data. In some cases, this led to adjustments, such as the dropping or adding of indicators, or the adjustment of weights and cut-offs. It is worth noting that some highly relevant dimensions and indicators (for example nutrition and child mortality) were not included in the present version of the measure due to a lack of adequate data.

2.1.3 Weights

The weights used in this report assign 1/3 of the MPI's total weight to each of the three core dimensions: education, health and living standards. Within education, different indicators are normally weighted equally with some adjustments to this nested weighted structure, which are explained as follows. Years of schooling is weighted at 1/6 (16.67%). The other 50% of the education domain focuses on school attendance, giving three quarters ¾ of the weight directly to child school attendance at 1/8 (12.5%), and the remaining weight to the quality of schooling, assessed by the indicator of educational quality at 1/24 (4.17%). Health indicators are also assigned different weights. Broadly speaking, access to health care accounts for 50% of the weights of this domain, while the other three indicators are equally weighted to comprise the remaining half, which reflect actions to prevent common health problems. Thus the first variable, access to health facilities/clinics is weighted at 1/6 (16.67%), while immunisation, ante-natal care, and assisted delivery are each assigned a weight of 1/18 (5.56%). Within the dimension of living standards, the indicators of water, sanitation, electricity, cooking fuel, assets, and land and livestock are each weighted at 1/21 (4.76%), while walls and overcrowding are weighted at 1/42 (2.38%) each because both represent different aspects of a housing component of living standards. Overall, the weights add up to 100%.

2.2 Alkire-Foster Methodology

The global MPI, developed by Alkire and Santos (2010, 2014) in collaboration with UNDP, first appeared in the 2010 Human Development Report. It represents one particular adaptation of the adjusted headcount ratio (M0) proposed by Alkire and Foster (2011) and elaborated by Alkire, Foster, Seth, Santos, Roche and Ballon (2015). This section outlines the methodology and its relevant properties used in the subsequent sections of this report to understand changes in multidimensional poverty in Pakistan.9

Sabina Alkire and James Foster's methodology for measuring multidimensional poverty identifies the extent of poverty by considering the intensity of deprivations which the poor suffer from (A), as well as the percentage of the population who are identified as poor (H). Mathematically, the MPI combines two aspects of poverty:

MPI = H x A

1. Incidence of poverty (H): the percentage of people who are identified as multidimensionally poor, or the poverty headcount.

   
   

2. Intensity of poverty (A): the average percentage of dimensions in which poor people are deprived.

2.2.1 The Multidimensional Poverty Index: An Adjusted Headcount Ratio

Within the adjusted headcount ratio methodology, a person is categorised as poor according to the MPI (“MPI poor”) in two steps. First, they are categorised as deprived or non-deprived in each indicator, by considering whether their achievements exceed a deprivation cut-off. The deprivation cut-off represents the minimum level of achievement someone must show to be considered non-deprived, in each MPI indicator. Based on this cut-off, a deprived individual receives a score of 1 while those who are not deprived receive a score of 0. These scores are multiplied by the weights previously assigned to each indicator, and then summed up to calculate the individual's weighted deprivation score across all indicators.

In the second step, second cut-off is used. This is the poverty cut- off (denoted as “k” in this study). In Pakistan's MPI it takes a value of 33.3%. This threshold is used to identify a person as multi- dimensionally poor. Hence, those individuals whose weighted deprivation scores are equal to or greater than 33.3% will be identified as multi-dimensionally poor. While those whose score does not exceeds 33.3% will be identified as non-por. These cut- off rates are described in more detail below.

All individuals categorised as MPI poor according to the dual cut- off methodology are then aggregated to calculate the poverty headcount ratio (denoted as H in the formula above). With respect to the calculation of the intensity of poverty (denoted as A in the formula above), the weighted deprivation scores of all individuals categorised as multi-dimensionally poor in a country's population are aggregated and then averaged.

Finally, the value of the headcount (H) and intensity (A) of poverty are multiplied to calculate the Multidimensional Poverty Index (MPI), as illustrated in the formula above.

2.2.2 Properties of the Multidimensional Poverty Index

This section outlines some of the features of the MPI that are especially useful for policy analysis. The first is that the MPI can be expressed as a product of two components: the share of the population who are multi- dimensionally poor, or the multidimensional headcount ratio (H), and the average deprivation scores among the poor, or the intensity of poverty (A).

This feature of the MPI has interesting policy implications for inter-temporal analysis. All reductions in the MPI occur because some deprivation experienced by a person categorised as 'poor' has been solved. A certain reduction in the MPI may manifest

either as a reduction of H (if removing a certain deprivation means that the person is no longer poor) or by reducing A (if removing this deprivation means that the person is still MPI poor but now experiences fewer deprivations). This difference cannot be understood merely by looking at the MPI's overall value. If a reduction in the MPI occurs merely by reducing the number of people who are marginally poor, then H decreases but A may not. On the other hand, if a reduction in the MPI occurs by reducing the deprivation experienced by the poorest of the poor, then A decreases, but H may not.

A second notable feature of the MPI is that, if the entire population is divided into m mutually exclusive and collectively exhaustive groups, the overall MPI can be expressed as a weighted average of the MPI values of m subgroups, where weights represent their respective population shares.

This feature, also known as“subgroup decomposability”, is useful for understanding the contribution of different subgroups to overall poverty levels.1 0 It is essential to note that the contribution of a subgroup to overall poverty depends both on the poverty level of that subgroup and on the subgroup's population share. Relevant population subgroups in Pakistan include populations in rural/urban areas, provinces and districts, as well as demographic groups.

Breaking down poverty in this way allows a closer analysis of multidimensional poverty, one which clearly reveals each indicator's contribution to poverty, as well as the changes in these contributions over time. It identifies the regions and groups which are the poorest, and determines whether they have 'caught up' or 'fallen behind' over time.

2.2.3 Poverty and Deprivation Cut-offs

As discussed above, thresholds are used to decide whether a person is multidimensionally poor, using the Alkire-Foster measurement framework. This involves: (a) a deprivation cut-off for each indicator, where a person is considered deprived in each indicator if their score falls below the cut-off; and (b) a cross- indicator cut-off (or poverty cut-off ), where a person is identified as poor if the weighted sum of their deprivations meets or exceeds the poverty cut-off.

For Pakistan's MPI, the poverty cut-off has been determined to be one-third of the indicators. Since the number of indicators considered is 15, a person who is deprived in at least one-third of these weighted indicators is considered multidimensionally poor. A person may be considered intensely poor if they are deprived in at least 50% of the indicators. We assess the robustness of Pakistan's MPI in terms of changes in the poverty cut-off and in the weights of indicators in the annex on robustness.

2.3 Data

The data used in this report to calculate Pakistan's national poverty measure is drawn from the Pakistan Social and Living Standards Measurement (PSLM) surveys for the years 2004/05, 2006/07, 2008/09, 2010/11, 2012/13 and 2014/15. 11

The PSLM surveys are designed to provide social and economic indicators in alternate years at both the provincial and district levels. The project was initiated in July 2004. Surveys have since been conducted every alternative year, with its latest wave undertaken in June 2015.

This survey tool has served as the main source of information for tracking Pakistan's progress on the Millennium Development

Goals (MDGs). This is largely because the surveys encompass questions on issues ranging from demographic characteristics to education, health, employment, household assets, household amenities, water supply and sanitation. In the years in which these surveys covered the provincial level, questionnaires also included information on households' main sources of income and consumption. To calculate Pakistan's MPI, survey waves representing the district level were selected to allow for greater regional disaggregation and comparisons.

The focal population of these surveys comprises populations in all urban and rural areas of Pakistan's four provinces, as well as the capital, Islamabad, and excluding military restricted areas. The sample size for the PSLM surveys at the district level is approximately 80,000 households. A two-stage stratified sample design was adopted in these surveys.

1. The report's detailed statistical methodology is provided as an Annex.

2. See Foster, Greer and Thorbecke (1984) for a discussion of this aspect of the MPI.

3. More details can be obtained at: http://www.pbs.gov.pk/content/pakistan-social-and-living-standards-measurement

10 Multidimensional Poverty in Pakistan

Chapter 2 Methodology | 11

Chapter 3 Main Results

National Uncensored Headcount Ratios

Pakistan’s National MPI: Key Results

The Composition of Poverty: Percentage Contributions of Each Indicator to the MPI

Chapter 3 Main Results

This chapter provides a detailed account of the national MPI results for Pakistan using data from the 2014/2015 PSLM survey. It discusses the current poverty outlook in the country at both the national and provincial/regional levels. We begin with a broad description of the indicators used by the MPI in Section 3.1. Thereafter, Section 3.2 presents Pakistan's national MPI results, while 3.3 unravels these results to reveal the composition of poverty by indicator.

3.1 National Uncensored Headcount Ratios

Uncensored headcount ratios represent the proportion of people who are deprived in each of the MPI's 15 indicators, irrespective of their poverty status. These are calculated without applying the second cut-off criterion used to categorise an individual as multidimensionally poor, i.e. whether he/she is deprived in one-third of the weighted indicators. Figure 3.1 presents these rates for 2014/15, allowing analysts to see, at a glance, the indicators with the highest and lowest levels of deprivation.

As this Figure shows, the greatest deprivations are found in cooking fuel (with 60.6% of the population deprived in this indicator), years of schooling (48.5%), assets (39.0%) and overcrowding (38.3%). The uncensored headcount ratios are lowest for the following indicators: households without a supply of electricity (6.4%), households in which a child was delivered without the assistance of trained personnel (8.2%), and households in which women who have given birth in the last three years did not receive ante-natal care (9.1%).

Figure 3.1

National Uncensored Headcount Ratios, 2014/15

Percentage of people who are deprived in each indicator, whether poor or not

contains a margin of error. Thus, the Table also presents a 95% confidence interval, which may be interpreted as indicating that we are 95% confident that Pakistan's true multidimensional poverty headcount ratio is between 37.3% and 40.2% of the population.

Table 3.1 Incidence, Intensity and Multidimensional Poverty Index (MPI), 2014/15 Survey Index Value Confidence Interval

(95%)

Source: Authors' calculations based on the 2014/15 PSLM survey

The average intensity of deprivation (A), which reflects the share of deprivations each poor person experiences on average, is 50.9%. That is, each poor person is, on average, deprived in almost half of the weighted indicators.

Since the MPI is the product of H and A, it yields a value of 0.197. This means that multidimensionally poor people in Pakistan experience 19.7% of the total deprivations that would be experienced if all people were deprived in all indicators.

Table 3.2 presents the headcount ratio (H) and the intensity of poverty (A) for urban and rural areas. As the Table reveals, poverty in rural areas is much higher than in urban areas and the difference is statistically significant. Although the intensity of deprivation is higher, overall, in rural Pakistan, this discrepancy is

60.6%

48.5%

18.5% 17.7%

32.4%

14.0%

9.1% 8.2%

18.5%

38.3%

School quality

Health facility

Immunisation

Electricity

Sanitation

Water

Cooking Fuel

Assets

6.4%

27.1%

10.9%

39.0%

28.0%

Years of schooling

School Attendance

Ante-natal care

Assisted delivery

Improved walls

Over crowding

Land & livestock (rural)

Education Health Living Standards Source: Authors' calculations based on the 2014/15 PSLM survey

3.2 Pakistan's National MPI: Key Results

Table 3.1 outlines Pakistan's MPI for 2014/15, as well as the value of its components: the proportion of people identified as multi- dimensionally poor (H) and the intensity of poverty (A). As the Table shows, the headcount ratio (H) of multidimensional poverty is 38.8%. Since this estimate is based on a sample, it

not nearly as great as the difference in the poverty headcount ratio between rural and urban areas. It is worth noting, moreover, that some two-thirds of Pakistan's population of more than 180 million live in rural areas.

Table 3.3 presents estimates for the MPI, H and A at the provincial and regional level, and Table 3.4 adds the confidence intervals.

Chapter 3 Main Results | 15

The broad pattern shows that among Pakistan's provinces, multidimensional poverty is highest in Balochistan and lowest in Punjab, whereas considering the standard errors, there is no

Table 3.4

Confidence Interval for Provincial Multidimensional Poverty

Multidimensional Poverty Index by National, Rural/Urban and Provincial/Regional Levels

significant difference between the MPI levels of Sindh and KP. It is also important to note that in all four provinces, poverty in rural

Province Value

Confidence Interval

(95%)

Figure 3.2

Multidimensional Poverty Index

0.600

0.500

0.400

0.415

0.295

0.394

0.482

0.337

Share (%)

(95%)

0.300

0.200

0.100

Overall

Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban

Overall

0.000

0.197

0.281

0.040

0.152

0.214

0.026

0.231

0.046

0.250

0.042

0.172

0.209

0.238

0.036

0.130

0.115

0.013

National Punjab Sindh KP Balochistan GB AJK FATA

Source: Authors' calculations based on data from the2012/13 and 2014/15 PSLM survey and the 2013/14 FATA Development Indicators Household Survey

Figure 3.3

Headcount (H)

Table 3.3

Multidimensional Poverty by Region, 2014/15

Province Value

Source: Authors' calculations based on data from the 2012/13 PLSM survey (for AJK and GB), the 2014/15 PSLM survey for other provinces and the 2013/14 FATA Development Indicators Household Survey

90.0%

80.0%

70.0%

60.0%

50.0%

40.0%

38.8%

54.6%

31.4%

43.7%

43.1%

75.5%

49.2%

57.8%

71.2%

84.6%

37.7%

43.2%

49.0%

28.1%

73.7%

Punjab

Sindh

Overall Rural Urban Overall Rural Urban

MPI

0.152

0.214

0.026

0.231

0.415

0.046

Incidence (H)

31.4%

43.7%

6.3%

43.1%

75.5%

10.6%

Intensity (A)

48.4%

48.9%

41.8%

53.5%

54.9%

43.4%

(FDIHS) for FATA.

30.0%

20.0%

10.0%

0.0%

9.4%

6.3%

10.6%

10.2%

7.9%

24.9%

3.1%

KP

Balochistan

AJK

Overall Rural Urban Overall Rural Urban Overall

0.250

0.295

0.042

0.394

0.482

0.172

0.115

49.2%

57.8%

10.2%

71.2%

84.6%

37.7%

24.9%

50.7%

51.1%

41.5%

55.3%

57.0%

45.7%

46.3%

Overall

Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban

Overall

National Punjab Sindh KP Balochistan GB AJK FATA

Source: Authors' calculations based on data from the 2012/13 and 2014/15 PSLM survey and the 2013/14 FATA Development Indicators Household Survey (FDIHS)

Figure 3.4

Intensity (A)

57.0%

GB

FATA

Rural Urban Overall Rural Urban

0.130

0.013

0.209

0.238

0.036

0.337

28.1%

3.1%

43.2%

49.0%

7.9%

73.7%

46.3%

41.0%

48.3%

48.3%

45.0%

45.8%

60.0%

50.9%

50.0%

40.0%

30.0%

51.6%

43.1%

48.4%

48.9%

41.8%

53.5%

54.9%

43.4%

50.7%

51.1%

41.5%

55.3%

45.7%

48.3%

48.3%

45.0%

46.3%

46.3%

41.0%

45.8%

Source: Authors' calculations based on data from the 2012/13 PSLM (for AJK and GB), the 2014/15 PSLM survey for other provinces and the 2013/14 FATA Development Indicators Household Survey (FDIHS) for FATA

20.0%

10.0%

Overall

Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban Overall Rural Urban

Overall

0.0%

12 The figures for FATA are reported using 2013/14 FATA Development Indicators Household Survey. However, the indicators differ somewhat from the national specifications due to missing data. The figures for GB and AJK have been calculated using 2012/13 PSLM Survey, owing to unavailability of data from the 2014/15 PSLM survey for these regions at the time of writing this report. While the values may not be strictly comparable, they nevertheless represent the most recent data available for each region.

16 Multidimensional Poverty in Pakistan

National Punjab Sindh KP Balochistan GB AJK FATA

Source: Authors' calculations based on data from the 2012/13 and 2014/15 PSLM survey and the 2013/14 FATA Development Indicators Household Survey (FDIHS)

Chapter 3 Main Results | 17

3.3The Composition of Poverty: Percentage Contributions of Each Indicator to the MPI

What deprivations create this level of poverty in Pakistan and how can they be reduced? To answer this question, this report takes a more in-depth view of multidimensional poverty by analysing the percentage which each of the 15 indicators contributes to Pakistan's MPI.

Figure 3.5 presents the weighted percentage contribution of each indicator to illustrate the composition of multidimensional poverty at the national level, and in rural and urban areas. It must be borne in mind that the weights assigned to most of the health and education indicators are higher than those assigned to the indicators concerning living standards. While all three core dimensions (education, health and living standards) are equally weighted, the indicators with greater weights in the spheres of education and health are expected to contribute relatively more to multidimensional poverty.

rural/urban levels, but there are fairly important differences, particularly in the relative contributions of the indicators pertaining to health and living standards. For instance, the indicator of school attendance contributes only 4.9% to total poverty in AJK, as opposed to nearly 12% in Sindh, nearly 13% in GB and 16% in FATA. On the other hand, deprivations in access to healthcare are highest in AJK, Punjab and KP contributing 21% to their respective MPIs, but falls to 8.1% and 8.9% in GB and FATA.

As demonstrated by Figure 3.7, the trends for FATA, GB and AJK vary slightly as opposed to national, provincial and rural/urban MPI decomposition. While deprivation in education is the highest contributing dimension for GB and FATA, standard of living contributes the most to poverty in AJK. At the indicator level, deprivation in cooking fuel is the third largest contributor for poverty in both GB and AJK. Secondly, the deprivation in child school attendance in AJK and access to health facilities in FATA and GB is significantly lower than all other provinces and regions.

Table 3.5

Percentage contribution of each indicator to MPI, by national and rural/urban

100% Years of schooling

Access to health facilities Assisted delivery Electricity

Cooking Fuel

At the national level, the indicators which contribute most to the MPI are years of schooling (29.7%), followed by access to health facilities (19.8%) and child school attendance (10.5%). At the dimensional level, deprivations in education are the largest contributor to the MPI (42.8%), followed by living standards (31.5%) and health (25.7%).

Figure 3.5 also reveals different profiles for urban and rural poverty. At the indicator level, the greatest contribution, in both urban and rural areas, derives from deprivation in years of schooling, access to health facilities, and child school attendance. In terms of dimensions, education is clearly the greatest contributor to multidimensional poverty in both areas, contributing almost 57% and 42%, respectively. It is followed by the dimension of living standards and, finally, the dimension of health. Notably, deprivation in health contributes almost 5.7% more to poverty in rural areas than it does in urban centres.

Figure 3.6 illustrates the break-down of multidimensional p ove r t y a t t h e p rov i n c i a l l e ve l. Th e co m p o s i t i o n o f multidimensional poverty is broadly similar across provinces and follows the same pattern as the MPI at the national and

80%

60%

40%

School Attendance Full immunisation Improved walls Sanitation

Assets

Educational quality Ante-natal care Overcrowding

Water

Land & Livestock

Table 3.5

Percentage Contributions of Indicators to MPI at the National and Provincial/Regional level

20%

0%

National

Urban Rural

Source: Authors' calculations based on the 2014/15 PSLM survey

Source: Authors' calculations based on the 2012/13 & 2014/15 PSLM survey and the 2013/14 FATA Development Indicators Household Survey

18 Multidimensional Poverty in Pakistan

Chapter 3 Main Results | 19

Table 3.6

Percentage contribution of each indicator to MPI, by province

Table 3.7

Percentage contribution of each indicator to MPI, by region

100%

80%

60%

Years of schooling Access to health facilities Assisted delivery

Electricity Cooking Fuel

School Attendance Full immunisation Improved walls Sanitation

Assets

Educational quality Ante-natal care Overcrowding

Water

Land & Livestock

100%

80%

60%

Years of schooling Access to health facilities Assisted delivery

Electricity Cooking Fuel

School Attendance Full immunisation Improved walls Sanitation

Assets

Educational quality Ante-natal care Overcrowding

Water

Land & Livestock

40%

40%

20%

20%

0%

Punjab

Sindh

KP Balochistan

0%

GB AJK

FATA

Source: Authors' calculations based on the 2014/15 PSLM survey

Source: Authors' calculations based on the 2012/13 PSLM survey and the 2013/14 FATA Development Indicators Household Survey

20 Multidimensional Poverty in Pakistan

Chapter 3 Main Results | 21

Chapter 4 Changes in

Multidimensional Poverty over Time Changes in National Uncensored

Headcount Ratios

Changes in Multidimensional Poverty Index and its Components Over Time

Changes in National Censored Headcount Ratios

Chapter 4 Changes in Multidimensional Poverty Over Time

A key question to ask is how poverty has changed over time. This chapter examines the evolution of multidimensional poverty in Pakistan between 2004/05 and 2014/15. Since annual PSLM survey data is only available for this time period, the MPI and its sub-indices were calculated using six waves of the PSLM surveys: 2004/05, 2006/07, 2008/09, 2010/11, 2012/13 and 2014/15. The

PSLM surveys for these six periods share a common survey design and questionnaire, allowing researchers to recreate exactly the same indicators for each year and to make robust comparisons across time.

4.1 Changes in National Uncensored Headcount Ratios

Figure 4.1 represents the proportion of people deprived in all of the MPI's indicators, irrespective of whether they can be categorised as multidimensionally poor or not. As this Figure reveals, improvements are evident in most of the indicators over time, in terms of reductions in the proportion of people deprived with respect to these indicators. The possession of assets, access to adequate sanitation and cooking fuel are the indicators which display the greatest absolute reduction in terms of uncensored headcount ratios (28.5%, 19.1% and 14%, respectively).

Figure 4.1

National Uncensored Headcount Ratios

Percentage of people who are deprived in each indicator, whether poor or not

80.0%

70.0%

60.0%

50.0%

40.0%

30.0%

20.0%

10.0%

Years of schooling

School Attendance

Educational quality

Health facilities

Immunisation

Ante-natal care

Assisted delivery

Improved walls

Overcrowding

Electricity

Sanitation

Water

Cooking Fuel

Assets

Land & livestock (rural)

0.0%

Education Health Living Standards

2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

Source: Authors' calculations based on various waves of the PSLM surveys

Chapter 4 Changes in Multidimensional Poverty Over Time | 25

4.2 Changes in the Multidimensional Poverty Index and its Components over time

Turning to the three key statistics of the MPI, Figures 4.2-4.4

Figure 4.4

National Intensity (A), 2004-2015

54.0%

53.5%

1. %

   Multidimensional Poverty Over Time in Punjab (2004-2015)

   
   

   Figure 4.5

   Punjab MPI, 2004-2015

   
   

   Figure 4.7

   Punjab Intensity, 2004-2015

   provide an overview of how the incidence (H) and intensity (A) of poverty, as well as the overall MPI, have changed over the years, using the four provinces for which data is available for each wave. It is evident that multidimensional poverty has declined gradually between 2004 and 2015 and that the reduction across the decade is statistically significant. The MPI dropped from

   0.292 in 2004/05 to 0.197 in 2014/15, while the headcount ratio

   (H) fell by over 16.4 percentage points, from 55.2% to 38.8%. Strikingly, however, the average deprivation share of the poor

   
   

   53.0%

   52.5%

   52.0%

   51.5%

   51.0%

   50.5%

   50.0%

   49.5%

   49.0%

   52.9%

   
   

   52.6%

   
   

   51.0%

   
   

   50.7% 50.9%

   0.300

   
   

   0.250

   
   

   0.200

   
   

   0.150

   
   

   0.100

   
   

   0.254

   
   

   0.239

   
   

   0.219

   
   

   0.188

   0.168

   
   

   0.152

   
   

   52.0%

   51.5%

   51.0%

   50.5%

   50.0%

   49.5%

   49.0%

   48.5%

   48.0%

   47.5%

   
   

   51.1% 51.4%

   
   

   50.6%

   
   

   49.5%

   
   

   48.5% 48.4%

   declined relatively little, from 52.9% to 50.9%. Nevertheless, on a

   positive note, Pakistan experienced statistically significant reductions in its MPI, H and A between 2004/5 and 2014/15 (see Table 4.1).13

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   For an in-depth look at how poverty has varied over time at a sub-

   0.050

   
   

   0.000

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   47.0%

   46.5%

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Table 4.1

   Cut-off (k=33%) 2004/05 (i)	MPI 0.292	Incidence (H) 55.2%	Intensity (A) 52.9%
   2006/07	0.281	52.5%	53.4%
   2008/09	0.260	49.3%	52.6%
   2010/11	0.228	44.7%	51.0%
   2012/13	0.207	40.8%	50.7%
   2014/15 (ii)	0.197	38.8%	50.9%
   Change 2004 (i) - 2015 (ii)	0.095***	0.164***	0.020***
   Combined SE	0.0052	0.0091	0.0025
   Hypothesis	18.16	17.99	8.08
   p-value	0.000	0.000	0.000

   Change overtime in Incidence, Intensity and the MPI, 2004-2015

   national level, the multidimensional poverty figures and their

   constituent components were also analysed separately for each province (see Figures 4.5 – 4.16). In all four provinces, the general trend is that of a decreasing MPI.

   
   

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Figure 4.6

   Punjab Headcount, 2004-2015

   
   

   60.0%

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   50.0%

   
   

   40.0%

   
   

   30.0%

   
   

   20.0%

   
   

   10.0%

   
   

   0.0%

   49.7%

   46.4%

   
   

   43.2%

   
   

   38.1%

   
   

   34.7% 31.4%

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   Note: *** 1% level of significance

   
   

   Figure 4.2

   National MPI, 2004-2015

   
   

   0.350

   
   

   55.2%

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Multidimensional Poverty Over Time in Sindh (2004-2015)

   
   

   Figure 4.8

   
   

   Figure 4.10

   
   

   0.300

   0.250

   0.292 0.281

   
   

   0.260

   0.228

   (2004-05)

   HEADCOUNT

   38.8%

   (2014-15)

   Sindh MPI, 2004-2015

   -

   Sindh Intensity, 2004-2015

   
   

   0.200

   0.150

   0.100

   0.050

   0.000

   0.207 0.197

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   52.9%

   (2004-05)

   X

   
   

   INTENSITY

   
   

   =

   
   

   50.9%

   (2014-15)

   0.350

   0.300

   0.250

   0.200

   0.150

   0.317

   
   

   0.302

   
   

   0.280

   
   

   0.252 0.236 0.231

   57.0%

   56.0%

   55.0%

   54.0%

   53.0%

   52.0%

   
   

   55.3%

   56.3%

   
   

   54.6%

   
   

   52.6% 53.0%

   
   

   53.5%

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   Figure 4.3

   National Incidence (H), 2004-2015

   
   

   0.292

   (2004-05)

   
   

   MPI

   
   

   0.197

   0.100

   0.050

   0.000

   51.0%

   50.0%

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   60.0%

   50.0%

   40.0%

   30.0%

   20.0%

   55.2% 52.5%

   
   

   49.3%

   
   

   44.7%

   
   

   40.8% 38.8%

   (2014-15)

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Figure 4.9

   Sindh Headcount, 2004-2015

   
   

   70.0%

   Source: Authors' calculations based on six waves of the PSLM surveys

   10.0%

   0.0%

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   60.0%

   50.0%

   40.0%

   57.3% 53.7%

   
   

   51.2% 48.0%

   44.6%

   
   

   43.1%

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   13 Since data for Gilgit-Baltistan (GB) and Azad Jammu & Kashmir (AJK) was only available for three waves of the PSLM surveys – 2006/07 (only GB), 2010/11 and 2012/13

   – all the national values reported for trend analysis do not include GB, AJK and the Federally Administered Tribal Areas (FATA). However, the difference in national values after including these regions is minimal and insignificant. Hence, their exclusion does not impact the overall analysis offered by this chapter.

   30.0%

   20.0%

   10.0%

   0.0%

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   26 Multidimensional Poverty in Pakistan

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Chapter 4 Changes in Multidimensional Poverty Over Time | 27

   

   
   

   Multidimensional Poverty Over Time in Khyber Pakhtunkhwa (2004-2015)

   To obtain a cumulative analysis across all provinces, Figures 4.17 and 4.18 also illustrate provincial level change in the MPI in absolute and relative terms, respectively. As these Figures show,

   consecutive survey waves shows no statistically significant changes in headcount ratio, and only one significant change in MPI between 2008 and 2010. However, both figures have

   Figure 4.11

   Khyber Pakhtunkhwa MPI, 2004-2015

   -

   Figure 4.13

   Khyber Pakhtunkhwa Intensity, 2004-2015

   53.5% 53.2%

   KP and Punjab demonstrate the greatest absolute reduction in their MPI between 2004 and 2015 (0.101 and 0.102 points of the Index, respectively). In addition, Punjab accounts for the highest

   significantly reduced over the period of 2004 to 2014.

   
   

   Table 4.2

   Statistical Significance of Change in Headcount for All Provinces

   0.400

   0.350 0.350

   
   

   53.0%

   53.0% 53.1%

   relative reduction (40.2%). On the other hand, although ?

   Balochistan experienced the slowest reduction in relative terms

   0.350

   0.300

   0.250

   0.200

   0.150

   0.100

   0.050

   0.000

   0.321

   
   

   0.280 0.249 0.250

   52.5%

   52.0%

   51.5%

   51.0%

   50.5%

   50.0%

   49.5%

   49.0%

   
   

   50.9% 50.8% 50.7%

   (17.7%), in absolute terms both Sindh and Balochistan

   experienced almost identical progress (a reduction of 0.086 and

   0.085 points of the Index, respectively). As noted above, it is worth recalling that although Punjab accounts for nearly 60% of Pakistan's total population, its incidence of multidimensional poverty was only slight higher than 30% in 2014/15 (31.4%). By contrast, Balochistan is home to a fewer than 5% of the country's population, 71% of whom are poor. Regrettably, this represents a potentially polarising case of horizontal inequality in which the

   Province

   
   

   Punjab

   Years

   
   

   2014-2012

   
   

   2012-2010

   2010-2008

   
   

   2008-2006

   2006-2004

   Change in Headcount (H)

   
   

   -0.03288**

   
   

   -0.0339**

   -0.0516**

   
   

   -0.03221**

   -0.03296**

   Change in MPI

   
   

   -0.01649**

   
   

   -0.01984**

   -0.03037**

   
   

   -0.02019**

   -0.01519*

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Figure 4.12

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   Source: Authors' calculations based on six waves of the PSLM surveys

   gap between Balochistan and other provinces is increasing.

   
   

   Figure 4.17

   Absolute change in MPI, 2004-2015

   2014-2004

   2014-2012

   2012-2010

   -0.18353**

   -0.01441

   -0.0344*

   -0.10208**

   -0.00523

   -0.01635

   
   

   
   

   70.0% 65.8% 56.1%

   60.0%

   
   

   60.5%

   55.0%

   
   

   0.000 (2004/05 - 0.254) (2004/05 - 0.317) (2004/05 - 0.350)

   -0.020

   
   

   (2004/05 -0.478)

   2008-2006

   2006-2004

   -0.02448

   -0.03568*

   -0.02282*

   -0.01452

   
   

   50.0%

   49.1% 49.2%

   -0.040

   2014-2004 -0.14131** -0.08606**

   40.0%

   30.0%

   20.0%

   10.0%

   -0.060

   -0.080

   -0.100

   -0.120

   
   

   -0.102

   
   

   -0.086

   
   

   -0.101

   
   

   -0.085

   2014-2012

   2012-2010

   2010-2008

   KP

   2008-2006

   0.00094

   -0.05931**

   -0.05431**

   -0.05638**

   0.00051

   -0.03129**

   -0.04052**

   -0.02955*

   0.0%

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   Figure 4.18

   2006-2004

   2014-2004

   0.00271

   -0.16635**

   -0.00007

   -0.10092**

   Source: Authors' calculations based on six waves of the PSLM surveys

   
   

   Multidimensional Poverty Over Time in Balochistan (2004-2015)

   Relative Change in MPI, 2004-2015

   
   

   Punjab

   
   

   Sindh

   
   

   KP Balochistan

   2014-2012

   2012-2010

   -0.00688

   -0.04015

   -0.01008

   -0.01143

   
   

   Figure 4.14

   Balochistan MPI, 2004-2015

   -

   
   

   Figure 4.16

   Balochistan Intensity, 2004-2015

   
   

   60.0%

   59.0%

   
   

   0.0%

   -5.0%

   -10.0%

   -15.0%

   (2004/05 - 0.254) (2004/05 - 0.317) (2004/05 - 0.350) (2004/05 - 0.478)

   
   

   Balochistan

   2010-2008

   2008-2006

   2006-2004

   -0.03012

   -0.00873

   -0.03584

   -0.04371**

   -0.01183

   -0.0075

   0.600

   
   

   0.500

   
   

   0.400

   
   

   0.478 0.471 0.459

   0.415 0.404 0.394

   59.0%

   58.0%

   57.0%

   
   

   57.4%

   58.2%

   
   

   56.2%

   -20.0%

   -25.0%

   -30.0%

   -35.0%

   
   

   -27.2% -28.8%

   
   

   -17.7%

   2014-2004 -0.12172** -0.08455**

   
   

   * Change is Statistically significant at 5% significance level

   
   

   0.300

   
   

   0.200

   
   

   0.100

   56.0%

   55.0%

   54.0%

   53.0%

   52.0%

   
   

   54.7%

   55.3%

   -40.0%

   -45.0% -40.2%

   
   

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   Table 4.2 reports changes in the incidence or headcount ratio (H)

   ** Change is Statistically Significant at 1% Significance level

   
   

   Source: Authors' calculation based on data from various waves of PSLM survey

   
   

   Poverty trends in rural and urban areas are depicted in Figures

   4.19 - 4.24. Rural areas experienced significant reductions in MPI

   0.000

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on six waves of the PSLM surveys

   and MPI over time across provinces. In particular, we look at the changes between consecutive waves of the survey (2004/05, 2006/07, 2008/09, 2010/11, 2012/13 and 2014/15), and between

   headcount ratio, which fell from 70.3% to 54.6%. That is, 15.6% of the population in rural areas emerged from poverty. In urban areas, poverty plummeted from 24% of the population to 9.4%,

   Source: Authors' calculations based on six waves of the PSLM surveys

   Figure 4.15

   Balochistan Headcount, 2004-2015

   
   

   86.0%

   the first and the last wave of the survey (2014 compared to 2004). The result suggest that both the headcount ratio (H) and MPI have significantly reduced in Punjab across all years. In Sindh, there has been a significant reduction in headcount between 2004-06, 2008-10 and 2010-12, and a significant overall

   signifying that 14.6% of the population living in urban areas

   'exited' poverty. While this may seem a similar result, it must be noted that the initial levels of poverty in rural and urban centres were quite different. Relative to their initial poverty headcount ratio, urban areas experienced a relative reduction of almost 64%

   84.0%

   82.0%

   80.0%

   78.0%

   76.0%

   74.0%

   72.0%

   70.0%

   68.0%

   66.0%

   64.0%

   		79.8% 78.9%
   						75.9%
   								71.9% 71.2

   83.4%

   
   

   %

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   reduction from 2004 to 2014. There has also been a significant decrease in MPI figures for Sindh, from 2008-10, and 2006-08, as well an overall reduction from 2004 to 2014.

   For KP, the changes have alternated between an increase and decrease in both MPI and the headcount ratio. We only observe a significant reduction in the headcount ratio and MPI between the years of 2006 and 2012, and a significant overall reduction between 2004 and 2014. For Balochistan, however, comparing

   in their MPI, compared to a relative reduction of 26% in rural

   areas. On the other hand, the intensity of poverty (A) has decreased only slightly and remains considerably higher in rural areas (51.6%) as compared to urban centres (43.1%).

   
   

   Source: Authors' calculations based on six waves of the PSLM surveys

   28 Multidimensional Poverty in Pakistan

   
   

   Chapter 4 Changes in Multidimensional Poverty Over Time | 29

   

   Multidimensional Poverty Over Time in Rural Areas (2004-2015)

   Figure 4.19

   Rural Areas' MPI, 2004-2015

   Figure 4.21

   Rural Areas' Intensity, 2004-2015

   
   

   4.3 Changes in National Censored Headcount Ratios

   
   

   four health indicators, increases in censored headcount ratios are

   0.400

   0.350

   
   

   0.379 0.380

   0.349

   0.312

   55.0%

   54.5%

   54.0% 5

   
   

   3.9%
   			53.6%
   					51.8% 51.4% 51.6%

   54.6%

   To understand how poverty has decreased in terms of the specific indicators driving its reduction, this section unpacks

   apparent at various points over the years.

   In terms of the indicators within the dimension of living

   0.300

   0.250

   0.200

   0.150

   0.100

   0.050

   0.000

   0.288

   0.281

   53.5%

   53.0%

   52.5%

   52.0%

   51.5%

   51.0%

   50.5%

   50.0%

   49.5%

   changes in the MPI according to each of the Index's component

   indicators. Figure 4.25 provides a refined view of what drove substantial reductions in Pakistan's multidimensional poverty over time. Censored headcount ratios, which measure the percentage of people who are “MPI poor” and who are deprived in a given indicator, are presented for each of the six periods covered by the PSLM surveys.

   Generally, trends indicate that censored headcount ratios have

   standards, substantial improvements are apparent with respect

   to assets, sanitation and cooking fuel. In all three of these indicators, censored headcount ratios declined gradually and substantially.

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   Figure 4.20

   Rural Areas' Headcount, 2004-2015

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   Source: Authors' calculations based on various waves of the PSLM surveys

   declined over time in each indicator, with the exception of immunisation (which had low initial levels of deprivation), and the ownership of land and livestock (where deprivations increased). Within the dimension of education, for instance, all three censored headcount ratios reveal significant reductions

   80.0%

   70.0%

   60.0%

   50.0%

   40.0%

   30.0%

   20.0%

   10.0%

   0.0%

   70.3% 69.5% 65.2%

   					60.2% 56.0% 54.6%

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   between 2004 and 2015. However, while the censored headcount ratio for educational quality has decreased during the period analysed, it witnessed a particularly sharp increase between 2006/07 and 2008/09. Similarly, within the dimension of health, although an overall reduction in censored headcount ratios took place, these did not follow a linear trend. Across the

   
   

   Figure 4.25

   National Censored Headcount Ratios, 2004-2015

   Percentage of people who are MPI poor and deprived in each indicator

   60.0%

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   Multidimensional Poverty Over Time in Urban Areas (2004-2015)

   
   

   Figure 4.22

   Urban Areas' MPI, 2004-2015

   
   

   Figure 4.24

   Urban Areas' Intensity, 2004-2015

   
   

   50.0%

   
   

   0.120

   
   

   0.100

   
   

   0.080

   
   

   0.060

   
   

   0.112

   
   

   0.088

   0.078

   
   

   0.054

   
   

   47.0%

   46.0%

   45.0%

   
   

   46.5%

   
   

   45.3% 45.2%

   
   

   43.1%

   
   

   40.0%

   
   

   0.040

   
   

   0.020

   
   

   0.000

   0.043

   0.040

   43.0%

   42.0%

   41.0%

   40.0%

   42.6% 42.6%

   
   

   20.0%

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   
   

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   10.0%

   
   

   Figure 4.23

   Urban Areas' Headcount, 2004-2015

   
   

   30.0%

   
   

   0.0%

   
   

   25.0%

   
   

   20.0%

   
   

   15.0%

   Years of schooling

   School Attendance

   Educational quality

   Health facilities

   Immunisation

   Ante-natal care

   Assisted delivery

   Improved walls

   Over crowding

   Electricity

   Sanitation

   Water

   Cooking Fuel

   Assets

   Land & livestock (rural)

   24.0%

   19.4%

   17.3%

   
   

   12.7%

   10.0%

   
   

   5.0%

   
   

   0.0%

   10.1%

   9.4%

   
   

   Education Health Living Standards

   
   

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   2004/05 2006/07 2008/09 2010/11 2012/13 2014/15

   Source: Authors' calculations based on various waves of the PSLM surveys

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   30 Multidimensional Poverty in Pakistan

   Chapter 4 Changes in Multidimensional Poverty Over Time | 31

   

   Figure 4.26 presents the absolute change in censored headcount ratios between 2004 and 2015, in percentage points, illustrating the percentage of the population previously considered poor and deprived in a particular indicator, that is now either non- poor, or non-deprived in that indicator. In addition to significant improvements with respect to assets, sanitation and cooking fuel, similarly impressive reductions are also evident in the censored headcount ratios of other indicators. These include years of schooling (14%) and child school attendance (9.2%). As these indicators are assigned substantial weights in the MPI, reductions in these spheres have driven significant changes in the national MPI. The only indicators which experienced a small but gradual increase in terms of their censored headcount ratios are land and livestock (rising by 0.8%) and immunisation (0.7%), both being of which are statistically significant.

   
   

   Figure 4.26

   Change in Censored Headcount Ratios, 2004 - 2015

   
   

   5.0%

   
   

   0.0%

   
   

   0.7%

   
   

   0.8%

   
   

   
   

   -5.0%

   
   

   -10.0%

   
   

   -9.2% -8.8% -8.4%

   
   

   -5.5% -6.5%

   
   

   -9.6%

   
   

   -7.9% -8.0%

   -3.4%

   
   

   -15.0%

   
   

   -20.0%

   
   

   -14.0%

   
   

   -17.4%

   
   

   -16.3%

   
   

   Years of schooling

   School Attendance

   Educational quality

   Health facilities

   Immunisation

   Ante-natal care

   Assisted delivery

   Improved walls

   Over crowding

   Electricity

   Sanitation

   Water

   Assets

   -25.0%

   
   

   Cooking Fuel

   -22.4%

   
   

   Land & livestock (rural)

   Education Health Living Standards

   
   

   Source: Authors' calculations based on data from various waves of the PSLM surveys

   
   

   32 Multidimensional Poverty in Pakistan

   

   

   
   

   Chapter 5 Multi-

   dimensional Poverty at District Level

   

   Chapter 5 Multidimensional Poverty at

   the District Level

   
   

   The headcount or incidence of poverty, as a key component of the MPI, is an excellent measure by which to determine the number of individuals who may be categorised as poor in any geographical region. To analyse poverty at a micro-level, this chapter presents the poverty headcount measure for all districts in Pakistan.

   Looking at MPI values across all districts, quite a divergent pattern appears. In the Figure 5.1, the starting level of MPI is plotted on the horizontal axis, with the highest poverty districts placed on the right. The absolute pace of poverty reduction is plotted vertically, with the best-performing districts appearing at the bottom of the graph as they are outrunning the rest interms of reducing MPI. Note that the zero value on the horizontal axis denotes no change in poverty, whereas positive valuesndicate an increase in poverty. The Figure illustrates that thepoorest district, Musakhel, which has data from all waves of PSLM sur vey witnessed the fastest reduc tion in MPI, demonstrating a positive and pro-poor trend. The relatively less poor districts suchas Islamabad, Lahore and Karachi experienced lower levels of absolute MPI reduction. However, a number of middle and highpoverty districts such as Ziarat, Killa Abdullah, and Chagai saw an increase in MPI values rather than a decrease during the period under analysis, while few other high poverty districts like Barkhan or Kohistan experienced only mild changes. These casesof poor districts increasing poverty makes the overall trend not clearly pro-poor, although there are certainly some positive cases within it.

   Figures 5.2 and 5.3 illustrate the absolute and relative change in headcount or incidence of poverty for all districts.14 As these Figures demonstrate, most districts have made significant progress in reducing their poverty headcount in both absolute and relative terms. While the MPI is the proper measure of multidimensional poverty, here we focus on the headcount ratio in order to present the simplest and most direct analysis for public dissemination.

   In absolute terms, the districts of Larkana, Attock, Malakand, T.T. Singh and Hyderabad have made the most progress, reducing poverty headcount ratio by more than 32 percentage points. In relative terms the best performers were the districts of Islamabad, Attock and Jhelum, followed by other big cities like Lahore, Karachi and Rawalpindi.

   On the other hand, some districts have experienced an increase in their poverty incidence. In absolute terms, the districts of Umerkot, Harnai, Panjgur, Killa Abdullah and Kashmore have witnessed the highest increase in incidence of poverty. Moreover, as revealed by the Figures below the same districts have also experienced the highest headcount increase in relative terms as well.

   Based on the index values for the latest year (2014/15), the five districts with the highest MPI are Killa Abdullah, Harnai, Barkhan, Kohistan and Ziarat. Most of these districts also have the highest levels of the incidence (headcount) and intensity of poverty in all of Pakistan. On the other hand, the six districts with the lowest index value are Islamabad, Lahore, Karachi, Rawalpindi, Jhelum and Attock. These districts also have the lowest poverty headcounts in the country.

   
   

   Figure 5.1

   Starting MPI value vs Absolute Reduction of MPI by District, 2004-2015

   
   

   14For most districts the relative headcount was calculated using the latest 2014/15 data and taking 2004/05 as a base year. However, the base year varies for those districts which were established after 2004 and are therefore not covered by the 2004/05 PSLM survey. Similarly, for two districts – Panjgur and Kech/Turbat data for 2014/15 was unavailable. As such, their headcount ratios for 2010/11 and 2012/13 were used as end points.

   
   

   Chapter 5 Multidimensional Poverty at the District Level | 37

   

   Figure 5.2

   Absolute Change in Headcount, 2004-2015

   
   

   -50.0% -40.0% -30.0% -20.0% -10.0% 0.0% 10.0% 20.0%

   Larkana (2004/05 - 81.3%) Attock (2004/05 - 43.0%)

   Malakand (2004/05 - 70.0%)

   T.T. Singh (2004/05 - 56.5%)

   Hyderabad (2004/05 - 57.8%) Kalat (2004/05 - 89.2%)

   Musakhel (2004/05 - 98.7%)

   Khuzdar (2004/05 - 88.8%)

   Dadu (2004/05 - 82.3%)

   Jhang (2004/05 - 71.7%)

   Haripur (2004/05 - 54.7%) Naushehro Feroze (2004/05 - 74.1%) Lodhran (2004/05 - 75.5%) Kasur (2004/05 - 48.1%)

   Pakpattan (2004/05 - 68.9%)

   Khairpur (2004/05 - 76.5%) Hafizabad (2004/05 - 57.2%) Chitral (2004/05 - 68.1%) Okara (2004/05 - 64.1%)

   Mansehra (2004/05 - 65.0%) Sahiwal (2004/05 - 54.8%) Narowal (2004/05 - 50.6%) Charsadda (2004/05 - 68.2%) Jhelum (2004/05 - 31.8%)

   Khanewal (2004/05 - 63.1%) Mardan (2004/05 - 56.8%)

   Nowshehra (2004/05 - 60.3%) Bhakkar (2004/05 - 74.4%) Loralai (2004/05 - 90.8%) Peshawar (2004/05 - 53.7%) Mandi Bahauddin (2004/05 - 52.4%) Lower Dir (2004/05 - 62.1%) Sialkot (2004/05 - 34.4%) Layyah (2004/05 - 65.9%)

   Multan (2004/05 - 55.9%)

   Kech/Turbat (2004/05 - 84.1%)

   Gujranwala (2004/05 - 32.3)

   Karak (2004/05 - 68.5%)

   Sarghodha (2004/05 - 53.2%)

   Mastung (2004/05 - 79.6%) Lakki Marwat (2004/05 - 80.0%) Sibi (2004/05 - 74.6%)

   Sheikhupura (2004/05 - 38.3%)

   Mianwali (2004/05 - 63.8%) Jamshoro (2008/09 - 72.4%) Swat (2004/05 - 71.6%)

   Faisalabad (2004/05 - 35.9%) Khushab (2004/05 - 56.9%) Nushki (2008/09 - 80.5%)

   Rawalpindi (2004/05 - 23.9%)

   Killa Saifullah (2004/05 - 95.0%) Awaran (2004/05 - 91.9%)

   Muzaffargarh (2004/05 - 79.4%)

   Swabi (2004/05 - 58.3%)

   Bolan/Kachhi (2004/05 - 87.4%)

   Nankana Sahib (2008/09 - 38.6%)

   Nasirabad (2004/05 - 90.8%)

   Vehari (2004/05 - 55.3%)

   Abbottabad (2004/05 - 46.2%)

   Sukkur (2004/05 - 52.8%)

   Tank (2004/05 - 84.2%)

   Rahim Yar Khan (2004/05 - 69.8%)

   Lasbela (2004/05 - 81.1%)

   Bannu (2004/05 - 71.5%)

   Kharan (2004/05 - 91.2%)

   Buner (2004/05 - 84.5%)

   Rajanpur (2004/05 - 77.1%)

   Bahawalpur (2004/05 - 65.1%) Upper Dir (2004/05 - 88.4%) D.G. Khan (2004/05 - 75.3%) Lahore (2004/05 - 15.9%)

   Gawadar (2004/05 - 72.3%) Kambar Shahdadkot (2008/09 - 83.4%) Hangu (2004/05 - 66.9%)

   Karachi (2004/05 - 15.4%) Kohat (2004/05 - 58.2%)

   Bahawalnagar (2004/05 - 60.6%)

   Chiniot (2010/11 - 52.5%) Washuk (2008/09 - 92.4%)

   Islamabad (2004/05 - 13.5%)

   Dera Bugti (2006/07 - 98.3%)

   Gujrat (2004/05 - 28.2%)

   Sanghar (2004/05 - 76.5%) Batagram (2004/05 - 84.9%)

   Nawabshah/ Shaheed Benazirabad (2004/05 - 69.0%) Chakwal (2004/05 - 22.4%)

   Zhob (2004/05 - 91.6%)

   Matiari (2008/09 - 70.7%)

   Jhal Magsi (2004/05 - 97.8%)

   Jaffarabad (2004/05 - 82.7%) Kohlu (2006/07 - 94.4%) Ghotki (2004/05 - 74.8%) Quetta (2004/05 - 53.5%)

   Jacobabad (2004/05 - 78.3%) Torgarh (2012/13 - 97.7%)

   Thatta (2004/05 - 84.1%) D.I. Khan (2004/05 - 71.0%)

   Shangla (2004/05 - 84.8%)

   Sherani (2010/11 - 92.9%) Badin (2004/05 - 76.7%)

   Kohistan (2004/05 - 96.9%)

   Mirpurkhas (2004/05 - 69.9%)

   Barkhan (2004/05 - 93.8%) Shikarpur (2004/05 - 59.8%) Chagai (2004/05 - 87.8%)

   Tharparkar (2004/05 - 85.0%) Pishin (2004/05 - 79.9%) Ziarat (2004/05 - 88.0%)

   Tando Allahyar (2008/09 - 64.2%)

   Tando Muhammad Khan (2008/09 - 75.0%) Kashmore (2008/09 - 71.3%) Killa Abdullah (2004/05 - 90.7%) Panjgur (2004/05 - 89.0%) Harnai (2010/11 - 86.2%)

   Umerkot (2010/11 - 75.9%)

   Figure 5.3

   Relative Change in Headcount, 2004-2015

   
   

   -80.0% -70.0% -60.0%

   
   

   -50.0% -40.0% -30.0% -20.0% -10.0% 0.0% 10.0% 20.0%

   Islamabad (2004/05 - 13.5%) Attock (2004/05 - 43.0%) Jhelum (2004/05 - 31.8%) Lahore (2004/05 - 15.9%)

   Karachi (2004/05 - 15.4%) Rawalpindi (2004/05 - 23.9%)

   Sialkot (2004/05 - 34.4%) T.T. Singh (2004/05 - 56.5%)

   Gujranwala (2004/05 - 32.3) Hyderabad (2004/05 - 57.8%)

   Haripur (2004/05 - 54.7%) Kasur (2004/05 - 48.1%)

   Larkana (2004/05 - 81.3%)

   Narowal (2004/05 - 50.6%)

   Malakand (2004/05 - 70.0%)

   Faisalabad (2004/05 - 35.9%)

   Sheikhupura (2004/05 - 38.3%)

   Sahiwal (2004/05 - 54.8%) Hafizabad (2004/05 - 57.2%) Chakwal (2004/05 - 22.4%) Jhang (2004/05 - 71.7%)

   Peshawar (2004/05 - 53.7%) Mardan (2004/05 - 56.8%)

   Mandi Bahauddin (2004/05 - 52.4%) Naushehro Feroze (2004/05 - 74.1%) Okara (2004/05 - 64.1%)

   Pakpattan (2004/05 - 68.9%)

   Nowshehra (2004/05 - 60.3%) Lodhran (2004/05 - 75.5%) Dadu (2004/05 - 82.3%)

   Mansehra (2004/05 - 65.0%) Khanewal (2004/05 - 63.1%) Chitral (2004/05 - 68.1%) Multan (2004/05 - 55.9%)

   Nankana Sahib (2008/09 - 38.6%) Kalat (2004/05 - 89.2%)

   Khuzdar (2004/05 - 88.8%) Gujrat (2004/05 - 28.2%)

   Charsadda (2004/05 - 68.2%) Sarghodha (2004/05 - 53.2%) Lower Dir (2004/05 - 62.1%) Khairpur (2004/05 - 76.5%) Musakhel (2004/05 - 98.7%) Layyah (2004/05 - 65.9%)

   Bhakkar (2004/05 - 74.4%) Khushab (2004/05 - 56.9%) Abbottabad (2004/05 - 46.2%) Karak (2004/05 - 68.5%)

   Mianwali (2004/05 - 63.8%) Sukkur (2004/05 - 52.8%) Swabi (2004/05 - 58.3%)

   Loralai (2004/05 - 90.8%) Vehari (2004/05 - 55.3%) Swat (2004/05 - 71.6%)

   Jamshoro (2008/09 - 72.4%) Sibi (2004/05 - 74.6%)

   Kech/Turbat (2004/05 - 84.1%) Mastung (2004/05 - 79.6%)

   Lakki Marwat (2004/05 - 80.0%) Nushki (2008/09 - 80.5%)

   Chiniot (2010/11 - 52.5%) Rahim Yar Khan (2004/05 - 69.8%) Bahawalpur (2004/05 - 65.1%)

   Muzaffargarh (2004/05 - 79.4%) Kohat (2004/05 - 58.2%) Bannu (2004/05 - 71.5%)

   Bahawalnagar (2004/05 - 60.6%)

   Hangu (2004/05 - 66.9%)

   Rajanpur (2004/05 - 77.1%) Killa Saifullah (2004/05 - 95.0%) Bolan/Kachhi (2004/05 - 87.4%) Awaran (2004/05 - 91.9%)

   Lasbela (2004/05 - 81.1%) Gawadar (2004/05 - 72.3%) Tank (2004/05 - 84.2%) D.G. Khan (2004/05 - 75.3%) Buner (2004/05 - 84.5%)

   Nasirabad (2004/05 - 90.8%) Kharan (2004/05 - 91.2%)

   Nawabshah/ Shaheed Benazirabad (2004/05 - 69.0%) Kambar Shahdadkot (2008/09 - 83.4%) Upper Dir (2004/05 - 88.4%)

   Quetta (2004/05 - 53.5%)

   Sanghar (2004/05 - 76.5%)

   Matiari (2008/09 - 70.7%) Batagram (2004/05 - 84.9%) Washuk (2008/09 - 92.4%)

   Dera Bugti (2006/07 - 98.3%)

   Ghotki (2004/05 - 74.8%) Zhob (2004/05 - 91.6%)

   Jaffarabad (2004/05 - 82.7%) Jacobabad (2004/05 - 78.3%) Jhal Magsi (2004/05 - 97.8%)

   Kohlu (2006/07 - 94.4%) D.I. Khan (2004/05 - 71.0%) Thatta (2004/05 - 84.1%)

   Torgarh (2012/13 - 97.7%)

   Shangla (2004/05 - 84.8%) Badin (2004/05 - 76.7%)

   Sherani (2010/11 - 92.9%)

   Mirpurkhas (2004/05 - 69.9%)

   Kohistan (2004/05 - 96.9%) Barkhan (2004/05 - 93.8%) Shikarpur (2004/05 - 59.8%) Chagai (2004/05 - 87.8%)

   Tharparkar (2004/05 - 85.0%) Ziarat (2004/05 - 88.0%) Pishin (2004/05 - 79.9%)

   Tando Muhammad Khan (2008/09 - 75.0%) Tando Allahyar (2008/09 - 64.2%)

   Kashmore (2008/09 - 71.3%) Killa Abdullah (2004/05 - 90.7%) Panjgur (2004/05 - 89.0%) Harnai (2010/11 - 86.2%)

   Umerkot (2010/11 - 75.9%)

   
   

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   Source: Authors' calculations based on various waves of the PSLM surveys

   
   

   38 Multidimensional Poverty in Pakistan

   Chapter 5 Multidimensional Poverty at the District Level | 39

   

   .,.

   0

   
   

   d

   a.

   3'

   "'

   "c;·

   
   

   ""0

   
   

   ;+

   '<

   s·

   
   

   ><C

   
   

   "

   
   

   Incidence of poverty by district:

   
   

   . Less than 10%

   
   

   -10%-19.9%

   D20%-29.9%

   
   

   . 30%-39.9%

   
   

   . 40%-49.9%

   D50%-59.9%

   D60%-69.9%

   
   

   . 70% and above

   f No data

   
   

   ,

   
   

   50 100 200 J<JO 400

   Kilometer'$

   -Pro,·ince Boundary District Boundary Line of Control

   
   

   Incidence of poverty by district:

   
   

   . Less than 10%

   
   

   -

   10%-19.9%

   D20%-29.9%

   
   

   . 30%-39.9%

   
   

   . 40%-49.9%

   D50%-59.9%

   D60%-69.9%

   
   

   . 70% and above

   
   

   No data

   
   

   n

   ::;;

   01

   !!:

   s

   "'

   c

   rt

   i'i

   3'

   0

   "'·

   "01

   ""0

   0

   <

   "'

   01

   
   

   "0'

   ::;;

   
   

   o;·

   s-.

   n

   [

   
   

   ...

   
   

   \

   N

   50 100 200 J<JO 400

   Kilometers

   • Province Boundary District Boundary Line of Control

     

     .,.

     IV

     
     

     d

     a.

     3'

     "'

     "c;·

     
     

     "'0

     
     

     ;+

     '<

     s·

     
     

     ><C

     
     

     "

     
     

     Incidence of poverty by district:

     
     

     . Less than 10%

     
     

     -

     10%-19.9%

     D20%-29.9%

     
     

     . 30%-39.9%

     
     

     . 40%-49.9%

     D50%-59.9%

     D60%-69.9%

     
     

     . 70% and above

     
     

     No data

     
     

     		_\
     50	100	N 200	300	400
     		Kikrneters

     • Province Boundary District Boundary Line of Control

       
       

       Incidence of poverty by district:

       
       

       . Less than 10%

       
       

       -

       10%-19.9%

       D20%-29.9%

       
       

       . 30%-39.9%

       
       

       . 40%-49.9%

       D50%-59.9%

       D60%-69.9%

       
       

       . 70% and above

       
       

       No data

       
       

       n

       ::;;

       01

       !!:

       s

       "'

       c

       rt

       i'i

       3'

       0

       "'·

       "01

       "'0

       0

       <

       "'

       01

       
       

       "0'

       ::;;

       
       

       o;·

       s-.

       n

       [

       .,.

       w

       
       

       N

       50 100 200 300 400

       -

       Kilometers

       
       

       Province Boundary District Boundary Line of Control

       

       
       

       .,.

       
       

       d

       a.

       3'

       "'

       "c;·

       
       

       "'0

       
       

       ;+

       '<

       s·

       
       

       ><C

       
       

       "

       
       

       Incidence of poverty by district:

       
       

       . Less than 10%

       
       

       -

       10%-19.9%

       D20%-29.9%

       
       

       . 30%-39.9%

       
       

       . 40%-49.9%

       D50%-59.9%

       D60%-69.9%

       
       

       . 70% and above

       
       

       No data

       
       

       N

       
       

       50 100 200 300 400

       Kikrneters

       
       

     • Province Boundary District Boundary Line of Control

Incidence of poverty by district:

. Less than 10%

-

10%-19.9%

D20%-29.9%

. 30%-39.9%

. 40%-49.9%

D50%-59.9%

D60%-69.9%

. 70% and above

No data

n

::;;

01

!!:

s

"'

c

rt

i'i

3'

0

"'·

"01

"'0

0

<

"'

01

"0'

::;;

o;·

s-.

n

[

.,.

U'l

N

50 100 200 300 400

-

Kilometers

Province Boundary District Boundary Line of Control

Chapter 6 Conclusion

Chapter 6 Conclusion

This report represents the endeavours of the Planning Commission of Pakistan to develop a different approach to measuring poverty in the country, in addition to conventional income-based poverty measures. Efforts to calculate the MPI were undertaken to complement existing measures which focus on income alone, as both measures offer important sources of information for public policy. In particular, Pakistan's national MPI can help monitor progress in terms of meeting the social and infrastructural goals outlined in its National Development Plan, Vision 2025.

Pakistan's national multidimensional poverty rate of 19.7% in 2014/15 varies from its income-based poverty rate of 29.5%, as estimated in 2013/14. This is because both measures use different criteria for determining poverty. Now that Pakistan has lower levels of extreme income poverty, it is appropriate to shine a light on the social situation through the lens of a Multidimensional Poverty Index. This is especially important as progress has been far slower on social indicators than it has with respect to economic ones. Thus, by using the MPI and identifying a higher percentage of people as poor, we are able to highlight them as worthy of policy attention.

To be identified as poor by the MPI, a person must be deprived in one-third of the Index's weighted indicators – that is, is between three and ten indicators, depending on their respective weights. It is worth stressing, however, that poor people are, on average, deprived in nearly 50% of the MPI's weighted indicators – that is, between five and thirteen indicators each. As such, not only is the poverty rate high, the MPI also reveals that significant deprivations are experienced by those identified as poor.

The MPI's value of 0.197 indicates that poor people in Pakistan experience 19.7% of the deprivations that would be experienced if all Pakistanis were deprived in all indicators. The greatest contribution to national poverty is made by indicators concerning deprivation in years of schooling (29.7%), access to health facilities (19.8%) and child school attendance (10.5%). If aggregated by dimensions, education contributes most to multidimensional poverty (42.8%), followed by the dimensions of living standards (31.5%) and health (25.7%).

Based on the report's findings and analysis, this concluding section presents a series of recommendations for policy makers and key stakeholders:

1. Use the MPI as a poverty measure which complements existing official measures, so as to offer a clearer outlook on poverty

   For the MPI to have an effective impact on policy design and constitute a useful tool for targeted interventions, it should be used alongside existing official income-based poverty measures. Regularly updated data on the MPI will help to determine which specific geographical regions, and which factors of deprivation, contribute most to national aggregate poverty. Monitoring changes in the MPI at the district, provincial and national levels will provide evidence to assess the success or failure of particular policies or initiatives.

   
   

2. Articulate the policy interface between Vision 2025 and the MPI

To catalyse the MPI's relevance for policy making, it would be useful to publish a succinct policy brief which itemises the connections between the MPI and Vision 2025 (as well as any recent commitments Pakistan has made with respect to the

SDGs) in greater detail than provided in this report. Ideally, this brief should elucidate the synergistic ways by which the MPI can reinforce and strengthen the implementation of Vision 2025 and help Pakistan progress towards meeting the SDGs. It should also elaborate how the MPI can help to monitor Pakistan's achievements in this regard.

3. Promote the use of the MPI for resource allocation

Following Pillar I of Pakistan's Vision 2025, the allocation of public sector resources should be informed by the MPI as well as by monetary poverty measures. Their complementary use in guiding policy will have the positive impact of improving sectoral policies across the country. A comparative analysis provided by the two measures will provide policy makers with a broader and more detailed outlook on poverty at the micro-level. This will serve as a better guide for budget allocations.

4. Issue provincial MPI reports

Drawing upon this report and its constituent data, summarised policy briefings should be prepared in local and regional languages. These should be shared with the Government, academia and other institutions operating in each region. This will support the provision of evidence-based policies in a devolved governance setting, while promoting targeted research and analysis. Such briefings will motivate key-players at the provincial level to become leaders and champions for reducing multidimensional poverty.

1. Promote the use of the MPI for district level policies

   
   

   District level policies should be informed by the composition of poverty in each district, as well as overall levels of poverty. This requires preparing district level reports on the MPI and issuing them to district offices. If such reports clearly highlight the contributing factors leading to poverty, district governments can improve their policies and implement initiatives targeting poverty and inequality in their regions.

   It is encouraging to note that poverty has decreased in most districts of Pakistan. While this commitment must be sustained, it is also important to conduct further analysis and research on each district to better understand the different situations they face and highlight successful cases.

   
   

   6. Include MPI variables in future surveys

   
   

   For strict comparability between different time periods, and to gauge progress over the years, all of the MPI's variables should be included in future surveys, especially the provincial PSLM surveys. Doing so will enable the MPI to be updated annually. This will increase its utility as a policy tool, since up-to-date information is vital for evidence-based policy making. As such, the lag between data collection and the MPI's release should be minimised.

   
   

   7. Include MPI variables in the next census

   
   

   Pakistan's next census should include as many MPI variables as possible, so as to comprehensively map poverty at the district level. This will help policy interventions at the grassroots level, spur local activism, and provide a crystal clear picture of multidimensional poverty in Pakistan.

   
   

   Chapter 6 Conclusion | 49

   

   1. Improve the national MPI's methodology and choice of indicators for future computations

      The consultations in different regions raised a plethora of suggestions regarding possible additional indicators. For example, although the PSLM surveys do not provide data on health functionings in general (such as nutrition and child mortality), efforts should be made to incorporate these issues in future surveys in order to improve computations of the national MPI. Finally, despite the difficulties in assessing education quality through surveys, innovative ways should be found to assess quality as an outcome-based indicator rather than as an input- based indicator, as it is treated at present.

      
      

   2. Promote future research

To understand the particular factors and policies which prompted reductions in poverty, as outlined in this report, it is recommended that further research be undertaken, particularly by the exceptionally strong community of scholars, economists and statisticians in Pakistan. This will bring to light specific districts that have successfully reduced multidimensional poverty in the shortest space of time, thereby allowing other districts to replicate policies by using these areas as a benchmark.

50 Multidimensional Poverty in Pakistan

Statistical

Annex

Annex 1: Reader's Guide to the Alkire-Foster Methodology

Annex 2: Robustness Analysis Annex 3: Statistical Tables

Annex 1 Reader's Guide to the

Alkire-Foster Methodology

Sabina Alkire and James Foster created a new method for measuring multidimensional poverty. It identifies who is poor by considering the intensity of the deprivations they suffer, and includes an aggregation method. Mathematically, the MPI combines two aspects of poverty:

MPI = H x A

1. I ncidence (H) of pover ty – the percentage of people who are

   multidimensionally poor, or the headcount of poverty.

   
   

2. Intensity of (A) of poverty – the average percentage of dimensions in which poor people are deprived.

.

14The meaning of the terms “dimension” and “indicator” differ slightly in Alkire and Foster (2011) and in Alkire and Santos (2010). In the former, no distinction is made between the two terms. In Alkire and Santos (2010), however, the term “dimension” refers to a pillar of well-being and may consist of several indicators.

Annex 1 Reader's Guide to the Alkire-Foster Methodology | 55

Annex 2 Robustness Analysis for MPI

1. For k=100%, the identification approach is referred to as the intersection approach. For 0<k< min {w ,...w }, it is referred to as

   i 1 d

   j

   the union approach (Atkinson, 2003). For min {w1,...wd}<k<1, it is referred to as the “dual cut-off approach” by Alkire and Foster,

   or more generally as the intermediate approach.

2. In the multidimensional context, there are two types of focus axioms. One is the deprivation focus, which requires that any increase in already non-deprived achievements should not affect the poverty measure. The other is the poverty focus, which requires that any increase in the achievements of non-poor persons should not affect the poverty measure. For more information, see Bourguignon and Chakravarty (2003) and Alkire and Foster (2011).

3. This feature is analogous to that of the Poverty Gap Ratio, which is similarly expressed as a product of the Headcount Ratio and the Average Income Gap Ratio among the poor.

4. Apablaza and Yalonetzky (2011) have shown that the change

5. See Foster, Greer and Thorbecke (1984) for a discussion of this property.

56 Multidimensional Poverty in Pakistan Annex 2 Robustness Analysis

| 57

0.1

58 Multidimensional Poverty in Pakistan

Annex 2 Robustness Analysis

| 59

60 Multidimensional Poverty in Pakistan Annex 2 Robustness Analysis

| 61