MOVING UP THE LADDER POVERTY REDUCTION AND SOCIAL MOBILITY IN NEPAL MOVING UP THE LADDER POVERTY REDUCTION AND SOCIAL MOBILITY IN NEPAL POVERTY AND EQUITY GLOBAL PRACTICE M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Vice President Annette Dixon Country Director Qimiao Fan Country Manager Takuya Kamata Practice Manager Benu Bidani Task Team Leader Sailesh Tiwari © The World Bank Group, Nepal All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the publisher. Published by: The World Bank Group Nepal Office P.O. Box 798 Yak and Yeti Hotel Complex Durbar Marg, Kathmandu, Nepal Tel.: 4236000 Fax: 4225112 Email: infonepal@worldbank.org www.worldbank.org/np www.facebook.com/WorldBankNepal Photos Courtesy Jaydev Poudyal/Stories of Nepal Design and Processed by: Spandan Design Communication Printed and bound in Nepal ii M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Foreword Policymakers will find this study insightful the leveraging of migration and diasporic networks because it disentangles hard evidence from in- as development resources. The pace of structural flammatory rhetoric at a time when the Nepali transformation determines the nature of growth. It state is going through a historic reconfiguration, is no longer just about reallocating activities across with heightened awareness about socio-ethnic sectors, but also within, increasingly shaped by new identity and political-economic opportunities. technologies and methods, such as the fragmenting Through a careful parsing of data, the report an- of production across borders. swers why poverty in Nepal has fallen sharply, how inequality has not increased, and yet vul- While concerns have emerged on whether the great nerability remains high. The report is one of the inventions of the past century are possibly irrepli- first to delve into the inequality of opportunities cable, or the degree to which a stagnating West can with the kind of rigor it employs. Evidence in fa- help lagging economies grow, or how threats like vor of substantial upward intergenerational mo- climate change restrict policy options and levers, bility reaffirms the importance of a functioning, there is much that a country like Nepal can achieve inclusive democracy, but also its limits in the simply through improved governance that mar- absence of consciously crafted public policies. shals people’s aspirations, ingenuity and wealth. The reduction of vulnerabilities remains a much We, however, need to be mindful of the challenges under-appreciated priority in Nepal. This needs that lie ahead and avoid mistakes of the past. Until to change. While catastrophic events grab all recently, inequality used to be explained away with the headlines, it is the dull, idiosyncratic disas- a set of “benign” causes such as skill bias in produc- ters that afflict everyday life. Almost a quarter tion and trade, disproportionate growth of the fi- of the country describes itself as belonging to nancial sector, and weakening bargaining power of the middle class, but remains a deeply insecure labor. Against the old wisdom that in the absence cohort. For a poor country, Nepal already has of impediments, inequality reflects talents and in place a generous welfare state, but the study choices, the revised view is that it can fan societal highlights the need for an overhaul of the social discord and stagnate economies, and that inequal- protection regime that must cut waste and abuse ity can be reduced with appropriate policies. This while deepening and expanding entitlements. is a hopeful message for countries like Nepal which The policy plea to equalize opportunities across are still at the starting point of the ascending half of the strained socio-ethnic and income fault lines the Kuznet’s Curve, as the report posits. of Nepal by improving the quality of services, increasing public investment and implementing I am confident that this study will influence the affirmative action is also apt. policy direction of the Government, and expect it to encourage universities, think tanks and commu- The report confirms that the most important nities of scholars to take up new lines of research economic challenge in Nepal is to enhance pro- enquiry. How do we expand opportunities, foster ductivity, both in agriculture and manufactur- mobility, and enhance the worth and dignity of ing, and to nurture our long-term sources of each Nepali life will continue to shape the arc of competitiveness in tradable services, including progress in our country for decades more. Dr. Swarnim Waglé Member of the National Planning Commission, Government of Nepal iii M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Acknowledgements This report is a synthesis piece that draws on four Parajuli, Saurav Dev Bhatta, Dhusyanth Raju, background papers prepared by a core team led by Yue Li, Uddhav Kumar Bhandari and Jayakrish- Sailesh Tiwari (Senior Economist, World Bank) na Upadhyay. Trishna Thapa, Gayatri Acharya, and comprising Hiroki Uematsu (Economist, and Jayadev Poudel supported the communica- World Bank), Akhmad Rizal Shidiq (Assistant Pro- tions effort while Cornelia Mirela Catuneanu fessor, University of Leiden), Carlos Felipe Balcazar and Sunita Kumari Yadav provided operational Salazar (Research Analyst, World Bank), Ram B. and logistical support. Khadka (Economist, Institute for Integrated De- velopment Studies) and Maheshwor Shrestha The World Bank Research Support Budget (DE- (Young Professional, World Bank). Mahesh Dahal CRSB) grant funded the Perceptions of Pover- (PhD candidate in Economics, Brown University) ty, Prosperity and Economic Mobility Survey provided excellent research assistance. which was an important source of data for the background research carried out for this work. We would like to thank the peer reviewers Gladys We would also like to thank United Kingdom’s Lopez Acevedo (Lead Economist, World Bank), Department for International Development Rinku Murgai (Lead Economist, World Bank) (DfID) for financial support that facilitated the and Luis-Felipe Lopez Calva (Lead Economist, qualitative research and dissemination activities World Bank) for helpful comments throughout of this project. the course of the task. For several useful conver- sations and suggestions that have strengthened Finally, we thank Takuya Kamata (Country various parts of the report, the team would also Manager, World Bank Nepal) and Benu Bidani like to thank Ambar Narayan, Tara Vishwanath, (Practice Manager, Poverty and Equity Global Iffath Sharif, Manohar Sharma, Aurelien Kruse, Practice of the World Bank) for their overall Rajib Upadhyay, Jasmine Rajbhandary, Dilip guidance and support throughout the process. iv M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L v M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Table of Contents Foreword--------------------------------------------------------------------------------------------------------------------------------------------------------iii Acknowledgements--------------------------------------------------------------------------------------------------------------------------------------------iv Overview--------------------------------------------------------------------------------------------------------------------------------------------------------1 I. Poverty and the Drivers of Poverty Reduction-------------------------------------------------------------17 Turbulent politics, weak growth---------------------------------------------------------------------------------------------------------------------17 Income poverty has declined ------------------------------------------------------------------------------------------------------------------------18 Remittances have played a part----------------------------------------------------------------------------------------------------------------------20 So have opportunities within Nepal for non-migrants-------------------------------------------------------------------------------------------21 II. Inequality and the Demand for Inclusion--------------------------------------------------------------------27 Demand for inclusion dominates political and economic discourse----------------------------------------------------------------------------27 Addressing inequality through public policy is difficult------------------------------------------------------------------------------------------30 But framing the debate along the lines of equity could bridge the divide---------------------------------------------------------------------31 Measuring inequality of opportunity---------------------------------------------------------------------------------------------------------------32 Deep inequality of basic opportunities-------------------------------------------------------------------------------------------------------------32 Opportunity gaps wider still, if quality taken into account--------------------------------------------------------------------------------------33 Large inequalities also in job opportunities--------------------------------------------------------------------------------------------------------35 III. Economic and Social Mobility----------------------------------------------------------------------------------41 Better educated than fathers, but not necessarily better employed-----------------------------------------------------------------------------41 Prospects of upward mobility have improved over time------------------------------------------------------------------------------------------41 But not for all social groups--------------------------------------------------------------------------------------------------------------------------42 Rural-urban migration, ticket to upward mobility------------------------------------------------------------------------------------------------42 Poverty transitions show significant movement even within a lifetime-------------------------------------------------------------------------42 Insecure and struggling middle class----------------------------------------------------------------------------------------------------------------47 IV. Migration and Other Pathways of Mobility-----------------------------------------------------------------51 Spatial extent of migration and consumer cities---------------------------------------------------------------------------------------------------51 Atypical structural transformation-------------------------------------------------------------------------------------------------------------------54 Too much of a good thing?---------------------------------------------------------------------------------------------------------------------------56 Migration may slow down...-------------------------------------------------------------------------------------------------------------------------58 ...but it will remain important-----------------------------------------------------------------------------------------------------------------------58 V. Directions for Policy- -------------------------------------------------------------------------------------------63 Equalizing Opportunity------------------------------------------------------------------------------------------------------------------------------63 Reducing Vulnerability--------------------------------------------------------------------------------------------------------------------------------65 Boosting Productivity---------------------------------------------------------------------------------------------------------------------------------68 References-----------------------------------------------------------------------------------------------------------------69 Annex-----------------------------------------------------------------------------------------------------------------------73 Annex 1: Human Opportunity Index (HOI) Methodology---------------------------------------------------------------------------------------------73 Annex 2: The synthetic panel methodology and application to Nepal---------------------------------------------------------------------------------75 vi M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L List of Table Table 1: Poverty Rates and Pace of Poverty Reduction in Nepal..........................................................................................18 Table 2: Gini Index of Consumption in Nepal in 1995-96 and 2010-11............................................................................28 Table 3: Upward and Downward Mobility in Education and Occupation, relative to fathers .............................................41 Table 4: Sources of Welfare Disparities between Various Geographic Areas of Nepal , 2011................................................53 List of Figures Figure 1: Where would Nepal be if it had grown at the rate at which some neighbors have grown?......................................17 Figure 2: Asset Ownership Among the Various Income Groups............................................................................................19 Figure 3: Reduced deprivations in non-income dimensions between 1995-96 and 2010-11.................................................20 Figure 4: Personal remittances Received in Nepal and Selected Countries.............................................................................21 Figure 5: Incidence of Remittance across the Consumption Distribution.............................................................................22 Figure 6: Drivers of Poverty Reduction between 1995-96 and 2010-11................................................................................23 Figure 7: Mean daily wages in agriculture and non-agriculture.............................................................................................25 Figure 8: Population and Population Growth in Nepal.........................................................................................................25 Figure 9: Frequency of Use of Some Key Words in the 1990 and 2015 Constitutions of Nepal, use per 1000 words............27 Figure 10: Gini Index by the Level of Development..............................................................................................................28 Figure 11: Welfare Shares by the top and bottom 10 percent..................................................................................................29 Figure 12: Growth Incidence Curve of Consumption and Income between 1995/96 and 2010/11........................................30 Figure 13: Circumstances Associated with Inequality of basic Opportunities..........................................................................33 Figure 14: Inequality in School Leaving Certificate Examination (2005) ...............................................................................35 Figure 15: Inequality in Passing Eighth Grade Student Assessment Examination, NASA 2013...............................................36 Figure 16: Inequality in Labor Market Opportunities, 2014...................................................................................................37 Figure 17: Upward Mobility by Age Cohorts..........................................................................................................................42 Figure 18: Upward Mobility by Caste/Ethnicity.....................................................................................................................43 Figure 19: Upward Mobility by Urban/Rural Migration Status...............................................................................................44 Figure 20: Episodes of movements in and out of poverty (% of population)...........................................................................45 Figure 21: The Poor, Vulnerable and the Middle Class in Nepal.............................................................................................46 Figure 22: Geographic Characteristics of the Poor, Vulnerable and the Middle Class, 2011....................................................46 Figure 23: The Poor, Vulnerable and the Middle Class in Nepal.............................................................................................47 Figure 24: The Proportion of Nepalis Struggling, Suffering and Thriving...............................................................................49 Figure 25: Increasing Spatial Extent of Migrant Population between 2001 and 2011.............................................................52 Figure 26: Spatial Variation in Changes in Real Wages in Various Regions of Nepal, 1995-2010............................................54 Figure 27: Primary source of employment of the economically active population in urban and rural areas in 1995 and 2010/11.....55 Figure 28: Sectoral composition of wage employment in urban and rural areas, 1995 and 2010.............................................56 Figure 29: Skill and gender composition of wage jobs, 1995 and 2010...................................................................................57 Figure 30: Gross Value Added, % of GDP..............................................................................................................................58 Figure 31: Monthly outflow of migrant workers from Nepal..................................................................................................59 Figure 32: Satisfaction with the availability of good jobs in the economy................................................................................61 vii M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L viii M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Overview Nepal is emerging from a series of crippling comparison to progress made by other South Asian Poverty declined shocks. Barely recovered from the devastation and neighbors. by an average loss wreaked by the earthquakes in April 2015, the country experienced a near total economic seizure Yet, despite overwhelming odds, Nepal made annual rate of between September 2015 and January 2016 as remarkable progress on improving living standards 2.2 percentage cross-border trade with India came to a halt. The and reducing poverty during this period. Absolute points between shortages of fuel, raw materials and other essential poverty rate declined by an average annual rate of 1995-96 and commodities caused prices to soar, businesses to 2.2 percentage points between 1995-96 and 2010- curtail operations and the economy to register 11 and currently stands at 25.2 percent. Though 2010-11 and the lowest growth experienced in the last fourteen there are some emerging signs that the pace of the currently stands years. decline is slowing down, the rate of poverty decline at 25.2 percent. seen in Nepal, particularly between 2003-04 and As devastating as these recent shocks have been, 2010-11, was among the fastest in the world. in a way, they are perhaps dramatic manifestations of how conflict and fragility have more generally The improvement in overall welfare has been impinged on Nepal’s development potential for accompanied by improvements in non-income the last two decades. The Maoist conflict between dimensions of well-being as well. Nepal met all 1996 and 2006, and the turbulent decade of of its MDGs and the Multidimensional Poverty transition that followed, took a heavy toll on Index (MPI) has also been in a steady decline. A the country’s investment climate and growth. significantly larger fraction of Nepalis today have Throughout this period, the economy operated better access to water and sanitation services, at a low level equilibrium, sputtering in fits and electricity and tele-connectivity than two decades starts around an average per capita growth rate ago. Fewer Nepali children are malnourished and of 2.5 percent, which is very low, especially in overall child and maternal mortality rates have FIGURE O.1: PERSONAL REMITTANCES RECEIVED IN NEPAL AND SELECTED COUNTRIES Source: World Development Indicators (World Bank, 2016), using personal remittances received as % of GDP, personal remittances received in current US dollars and the total population. 1 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L This report declined as well. Households with school- compare to the rest of the world? Has inequality aged children (6 to 15 years old) out of school been increasing? What kinds of inequality do deepens the declined from 55 percent in 1995-96 to 16 the Nepali people care about the most and understanding percent in 2010-11. This is presumably due to what can be done about it? Finally, the report of the key improvements in accessibility to and availability also presents some novel insights on social and processes that of schools, as households with children spending economic mobility experienced by Nepalis more than 20 minutes to reach primary school across generations as well as over a lifetime. have driven declined from 84 percent to 33 percent during Have Nepalis today done better than their improvements in the same period. But much of it is also likely due parents? How does this mobility experience vary living standards to increased ability to afford schools, including across the population? Who does better and who and identifies private schools. remains stuck? Are Nepalis today better off than they were five years ago? How much movement some of the How can these recent gains be cemented and along the proverbial socio-economic ladder challenges that made sustainable and how can progress be accel- is there in the society? What characteristics lie ahead. erated to improve the living standards of those are associated with these movements? That is, who remain in chronic poverty? Going forward, who moves up and who moves down? Has the this is one of the most important challenges fac- success in reducing poverty over the last two ing policymakers in Nepal today. This report decades translated into commensurate success in attempts to contribute to this debate by deep- building a middle class? How large is the Nepali ening the understanding of the key elements of middle class and has it been growing over this the processes that have driven improvements in period? living standards and identify some of the chal- lenges that lie ahead. Drivers of Poverty Reduction The most commonly cited driver of the progress The report also performs a careful analysis of on living standards and poverty reduction is the the levels as well as the trends in inequality over drastic increase in remittances sent by hundreds last two decades in an attempt to understand of thousands of Nepalis working overseas since the economic underpinnings of the demand the late 1990’s. Not only did more Nepali house- for inclusion that has fundamentally shaped holds start receiving remittances, the amount of Nepal’s contemporary socio-political landscape. remittances received by the remittance receiving How high is inequality in Nepal? How does it households also increased over the period. The size of remittances increased from 1.3 percent FIGURE 0.2: DRIVERS OF POVERTY REDUCTION BETWEEN 1995-96 AND 2010-11 of GDP in 1995 to 23 percent in 2010. Today remittances are around 29 percent of GDP. Remittances account for 16 percent of average household income, up from 6 percent in 1995- 96. Given this growth, it is entirely plausible for households with migrants to have benefited from remittances and realized improvements in living standards. But households without migrants have also benefited from opportunities created by the changes in the structure of the domestic econ- omy brought about, indirectly, by migration and remittances. Our decomposition exercises show that increase in labor income - or income derived from wage and non-wage employment within Nepal – accounts for 52 percent of the observed change in poverty nationally. In com- parison, the direct effect of remittances accounts Source: World Bank staff calculations based on data from NLSS 1995-96 and 2010-11 Note: Capital income and public transfers are under residual. 2 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L for about 27 percent of the overall poverty re- in the demand for non-farm labor which was The demand duction nationally and 33 percent of the reduc- in turn driven by economic diversification and for inclusion tion in rural areas. Further, breaking down the increase in non-farm activities brought about sources of income between farm and non-farm by remittance induced growth in incomes and that pervades activities, the growth in non-farm activities has the associated changes in demand for goods political and been more important for the observed reduc- and services. Non-wage employment in off economic tion in poverty than farm incomes. Changes in farm sectors has also increased suggesting that discourse in the demographic structure of the household – opportunities created outside of agriculture, which captures, among others, the number of particularly in rural areas has been the key ticket Nepal today is a household members that participate in the labor out of poverty in the preceding two decades. consequence of market relative to the number of dependents deep structural they have – also appears to have been important Inequality and the Demand inequalities for poverty reduction. for Inclusion The demand for inclusion that pervades politi- perceived by The role of labor income in the observed cal and economic discourse in Nepal today is a some segments of reduction of poverty can be explained, in part, consequence of deep structural inequalities per- the population. by rising wages. Real wages grew by about 4.2 ceived by some segments of the population. The percent per year between 1995-96 and 2010-11 10-year long Maoist conflict, abolition of a 250 with agricultural wages growing slightly faster year old monarchy, and the wholesale reconfigu- than non-agricultural wages. The growth in ration of the state structure, are all direct or in- agricultural wages was driven by the tightening direct expressions of a demand for a new, more of the supply of labor due to overseas migration inclusive social contract. Even though there has of working age males and increasing non-farm been some progress on this in the newly pro- activity. The increase in non-agricultural wages, mulgated constitution, lingering dissatisfaction on the other hand, was largely due to the growth continues to fuel unrest among some groups. FIGURE 0.3: GINI INDEX BY THE LEVEL OF DEVELOPMENT East Asia & Pacific Europe & Central Asia 60 Latin America & Caribbean Middle East & North Africa North America 50 South Asia Sub-Saharan Africa 40 Gini Index 30 20 Nepal (2010) Gini: 32.75 GNI per capita (Atlas Method): 540 10 10 100 1,000 10,000 100,000 GNI per capita (Atlas Method) on Log Scale Source: World Development Indicators (World Bank, 2016), using all countries for which Gini Index is available between 2009 and 2013. 3 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L But despite obvious implications it has for the for the historically disadvantaged groups. But risk of protracted conflict and political instabil- while a segment of the population may see this ity, which would weaken investment climate as an effort (if only in intention) in the right and hurt growth further, existing debates on this direction, another may object on the grounds of issue appear to provide very little basis on which violation of principles of meritocracy and per- to forge a constructive path forward. verse incentives these may create. Yet another group may acknowledge the need for redistribu- One challenge – at least on the part of the inclu- tion but disagree on the principle that guides the Leveling the sion agenda that overlaps with economic out- decision on who the beneficiary of such policies comes – is that even though there are persistent should be. Expressed in the form of a familiar re- playing field for gaps between urban and rural areas, between frain in Nepal, that “class and not caste” should all could move the regions and the various social groups, conven- determine who benefits from redistribution, this debate forward tional measures of inequality, such as the Gini point of view contends that any affirmative ac- because what index of consumption, show overall inequality tion policies should be based on economic dis- in Nepal to be fairly low and stable over the long advantage, as opposed to social disadvantage. is a minimum run. The Gini index for consumption for Nepal criterion for was 0.33 in 2010/11, which is at the lower end Equity, not Equality fairness and of global range seen for this number. This mea- But framing the debate along the lines of equity justice for the left sure of inequality also remained virtually un- as opposed to equality could provide a way for- changed between 1995-96 and 2010-11 both at ward and that is the central thrust of this report. happens also to be the national level and in rural areas. In the urban A society may have disagreements about the the foundation for areas, there has actually been a sharp decrease. sources of past injustices and what to do about meritocracy for them. But looking to the future, everybody can the right. Inequality in Nepal measured this way is not agree that opportunities for a better life for Ne- only low in absolute terms, it is low also rela- pali children should not depend on which part tive to the level of economic development. Ne- of the country they happen to be born in, or to pal’s Gini index is one of the lowest among the what caste /ethnicity group their parents hap- countries with a similar level of GNI per capita, pen to belong to or what language they speak or most of which are countries in Sub-Saharan Af- other such characteristics essentially determined rica, and well within the range of more advanced by the lottery of birth. This requires making a countries in Europe and Central Asia. If one is distinction between inequality of opportunities to imagine a cross-country Kuznets curve over and inequality of outcomes. this plot, it would appear that Nepal is at very early stages of development and that the engine The focus of development debate is often on in- of growth has not started cranking up to the ex- equality of outcomes, recasting the same discus- tent where this growth acceleration would begin sion around inequality of opportunities has an to generate a widening consumption distribu- intuitive appeal and could potentially transcend tion. deep ideological differences and effect action. An inclusive society is necessarily also an equal The other part of the challenge is that addressing opportunity society (even though the converse inequality through ex-post redistributive poli- many not be true). So working towards leveling cies is difficult, particularly at Nepal’s level of de- the playing field for all could move the debate velopment. Redistribution through progressive forward because what is a minimum criterion tax policies (in the space of income) or through for fairness and justice for the left happens also quotas and other forms of affirmative action in to be the foundation for meritocracy for the higher education (e.g. medical and engineering right. colleges) and employment for the disadvantaged groups are typical policy measures taken to re- In a society in which opportunities are equal, dress inequality. The newly promulgated consti- a girl (let’s call her Goma) born to a poor and tution and the recent amendment to the Civil illiterate Dalit couple in rural Kalikot should Service Act contain a number of provisions for have an equal shot at becoming a doctor, or an reservations and quotas to expand opportunities engineer as a boy (let’s call him Avidit) born in 4 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L BOX 1: OPPORTUNITY FOR A BETTER LIFE FOR GOMA, CHAMPA AND AVIDIT Goma is a girl, born in rural Kalikot. Her parents are more likely to start primary school late, drop out of illiterate, belong to the Dalit community and are in school, be stunted and grow up with limited access the bottom 20 percent of Nepal’s wealth distribution. to basic amenities such as clean water, sanitation Champa is also a girl born to a household otherwise and electricity than Avidit. Even in the 40 percent very similar to Goma’s. But Champa’s parents are chance that Goma and Champa complete primary from a village in Siraha. Avidit is a boy born to an school and go on into secondary and tertiary upper caste household in urban Kathmandu. Both his education, the quality of the education they receive parents have a university education and come from is likely to be vastly poorer than what Avidit will affluent backgrounds. receive in Kathmandu. In a society where opportunities are equally available Society may disagree about what to do about the for children of all socio-economic backgrounds, social and economic processes that led to their Goma, Avidit and Champa would all have equal parents’ being where they are in the current socio- odds becoming doctors, or engineers or successful economic ladder of Nepal. But everyone will agree entrepreneurs and businessmen. But in Nepal, the that position should not determine the children’s life trajectory of these children begins to diverge life chances. Efforts to equalize basic opportunities very early in life. Goma – whose life chances appear for these children can be a critical part of Nepal’s marginally better than Champa’s – is significantly inclusion agenda. 100 100 95 96 95 90 87 85 84 84 84 80 80 70 69 70 60 53 50 47 48 48 44 39 40 36 31 30 20 9 10 6 6 5 0 Started primary Finished primary Fully immunized Exclusively breastfed Not stunted Clean water Adequate Electricity on time on time sanitation Avidit Goma Champa Source: World Bank Staff estimates based on data from DHS (2011) a well-educated, affluent and upper caste house- that basic opportunities such as starting primary hold in urban Kathmandu. And so should a girl school on time, finishing primary school on (let’s call her Champa) born to parents of other- time, escaping the scourge of chronic malnutri- wise similar characteristics as Goma but living in tion in early childhood, having clean water to rural Sarlahi. But how equal/unequal are basic drink, clean and adequate sanitation facilities opportunities for children such as Goma, Avidit and electricity at home appear to be far from and Champa in Nepal? universal and unevenly distributed even among those who have it. The size of the inequality in Addressing this question empirically would be opportunities we report implies a significant di- difficult without some methodological advances vergence in life trajectories of Nepali children that have been made in the literature in this field based simply on the characteristics of the par- in the last decade. Applying this methodology to ents. There is quite a bit of variation in the cir- a variety of datasets in Nepal, this report finds cumstances that contribute to these inequalities, 5 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L but parental wealth, place of birth/residence expressions as opposed to those that require the (whether its urban or rural, in the mountains, solving of numerical and logical problems. This hills or terai, or if it is the central and eastern supports the idea that equalizing opportunities region or the mid and far western region) and for children in Nepal today is every bit about other parental characteristics (such as age, gen- leveling playing fields along the geographic and der and education) appear to account for most economic divide, as it is about closing the social of the observed inequality. divide. A mobile Inequalities are further pronounced when the This is not to say that there are no gaps in op- underlying quality of the opportunities are tak- portunities by caste/ethnicity of children. Even society is also en into account. A child may live in a household though the direct effects are small, the indirect a meritocratic connected to a tap that is never serviced. She effects could be larger. Caste/ethnicity status of one and such a may go to schools but not learn the material ad- parents could be correlated with other charac- society is better equately enough to progress through the grades teristics like education level or wealth status. For in the timely manner and eventually acquire the instance, a large part of why a Dalit couple is able to mobilize skills rewarded in the labor market. An ideal poor or uneducated in the first place could be be- and utilize measure of opportunity should also account cause of limited opportunities they had in their talents from all not only access but also for differences in qual- childhood. But distinguishing the effect on their segments of the ity. Data limits the extent to which we can fully children’s opportunities of being from the Dalit explore this, but using performance in national community (direct effect) versus being from a population. examinations such as the SLC or the National poor family with uneducated parents who are Assessment of Student Achievement (NASA) potentially poor and uneducated because they administered to eight-graders, we find the odds are Dalits (indirect effect) is difficult with the data of passing to be skewed in favor of children with we are using here. Evidence on social differences favorable birth circumstances: children whose in intergenerational mobility suggests that these parents can afford to send them to private indirect effects could be quite large. schools, whose parents are themselves educated and who happen to live in certain parts of the Intergenerational Mobility country more so than others. Inequality of opportunities is intricately linked to intergenerational mobility as both encapsu- Relative to some of the other drivers such as late the idea of origin independence. In an equal parental wealth and education and region of opportunity society life chances do not depend residence, the caste/ethnicity status of chil- on the social and economic identity of one’s par- dren accounts for small shares of total inequal- ent. In a society with high mobility across gener- ity. Interestingly, the language spoken at home ations, the economic and social position realized (whether or not it is Nepali) commands a non- by sons and daughters are independent of the trivial amount in the inequality of passing the economic and social positions of their parents. Nepali language (10 percent) and science (6 In both societies, who ones’ parents are is irrel- percent), but not for mathematics. Given that evant to the success one can expect to achieve. the medium of instruction in public schools is almost always Nepali, this suggests the pos- Sensitivity to inequality gets amplified in soci- sibility of non-native Nepali speakers having a eties that lack mobility. Conversely, societies disadvantage in subjects that require free form with high degree of mobility are able to heal the TABLE 0.1. UPWARD AND DOWNWARD MOBILITY IN EDUCATION AND OCCUPATION, RELATIVE TO FATHERS Education Occupation (% of population > 30 yrs.) (% of population) Up 61.9 43.6 Same 29.8 46.8 Down 8.3 9.6 Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and Economic Mobility in Nepal, 2014 6 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 0.4: WHAT CONTRIBUTES TO INEQUALITY OF BASIC OPPORTUNITIES FOR CHILDREN IN NEPAL? (a) Basic Opportunities (%) (b) Passing the eighth grade exam (%) Source: World Bank Staff calculations. Panel (a) uses data from DHS 2011; and Panel (b) is based on data from NASA Survey (2013), Ministry of Education, Government of Nepal wounds of structural inequalities. In an inter- being in a better occupation, we see that about generational sense, when children can aspire to 61.9 percent Nepalis (above the age of 30) expe- achieve and indeed achieve levels of education, rienced upward mobility in education. In com- jobs, and living standards that are materially dif- parison, about 29.8 percent stayed at the levels ferent from the levels enjoyed by their parents, of their fathers and about 8.3 percent had levels inequality begins to decline over time. Mobil- of education lower than their fathers. There is ity is important also from the perspective of less mobility in occupation than in education: dynamic efficiency. A mobile society is necessar- 47 percent of the overall population remains in ily also a meritocratic one and such a society is the same occupation as their father. But about better able to mobilize and utilize these talents 43.6 percent of the population realized upward by allowing people from all segments of society mobility. (including the disadvantaged segments) to con- tribute to the growth process. Prospects for upward mobility have been in- creasing in Nepal for both education and oc- Studying intergenerational mobility in the space cupation. Point estimates suggest that while the of income requires data that is hard to get, not likelihood of being better educated than one’s just in Nepal but also in most other developed father was below 50 percent for those born be- countries. But we attempt to make some prog- fore 1965, the number is closer to 70 percent ress on this using educational attainment and for the cohort born between 1985 and 1995. occupation, two strong proximate correlates of The pattern is similar for occupation as well income. Defining upward mobility as attaining except there is a discontinuous jump (roughly a higher level of education than one’s father or 10 percentage points) in predicted probability 7 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 0.5: UPWARD MOBILITY BY AGE COHORTS of upward mobility for the 1975-1985 birth-co- hort in comparison to the earlier cohorts. One (a) Better educated than father possible explanation for this is the fact that in contrast to the earlier cohorts, this group of chil- dren would have been in school around the time .7 Nepal’s economy started liberalizing (1990- 1994) and thus may have been in a favorable .65 position to seize the opportunities created by the Predicted probability consequences of this liberalization. In any case, younger cohorts have been more upwardly mo- .6 bile than the cohorts born before 1975 who ap- pear to have a low steady state transition prob- .55 ability of around 33-34 percent of doing better than their father. .5 However, prospects of upward mobility vary among the different social groups and some have .45 Before 1965 done better than others. In particular, Newars, followed closely by hill Brahmins and Chhetris (b) Better occupational status than father have the best odds of surpassing their father’s education and occupation levels. In contrast, .55 Nepalis in Muslim and other minority caste categories followed by Terai Brahmin/Middle Caste, and Dalit categories have the lowest odds .5 of upward mobility for education as well as oc- cupation. Predicted probability .45 Movements along the Ladder – The Vulnerable and the Middle Class .4 Recent events (earthquakes and the trade dis- ruptions) have put a global spotlight on the fragility of lives and livelihoods of Nepalis. But .35 even beyond these major events, a majority of Nepalis grapple with smaller and more mun- .3 dane disasters on a routine basis. As important Before 1965 as it is for the Government to respond to the Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and specific emergency of the earthquake related re- Economic Mobility in Nepal, 2014 construction and rehabilitation, this should also Note: The lines around the bars show 95% confidence interval and the x-axis denotes cohorts based on year of birth. serve as an opportunity to galvanize support for the design of systems that can respond in a more agile and efficient manner, not just to catastroph- ic disasters of similar scale (if need be) but also to smaller but more frequent shocks many Ne- palis endure. These “dull disasters” take a heavy toll on poor households by limiting their ability to maintain investments in the human capital of children and depleting their savings and assets. Inadequate protection from these shocks limits the ability of households to take risks and make productivity enhancing investments in both ag- riculture and non-agricultural activities. 8 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L One statistic that makes these vulnerabilities ap- FIGURE 0.6: UPWARD MOBILITY BY CASTE/ETHNICITY parent is the statistic on the movements in and out of poverty. The commonly reported poverty (a) Better educated than father trend (poverty declined from x percent to y per- .8 cent) often obscures two crucial and confound- ing phenomena of people moving out of poverty .7 and falling into poverty. These two are distinct processes with different implications for policy. Predicted probability Our analysis shows that between 1995-96 and .6 2003-04, 26 percent of the population moved out of poverty while 17 percent moved in the opposite direction and fell into poverty. Like- .5 wise, between 2003-04 and 2010-11, 29 per- cent moved out of poverty while 14 percent fell .4 back in. The numbers essentially imply that for every 2 Nepalis that escaped poverty in 2010- 11, one slid back into poverty. .3 Hill Brah. Hill Chhe. Terai B&MC Dalit Newar Hill Jana. Terai Jana. Musl.& Oth Health and weather shocks often represent the most common factors leading to a downward (b) Better occupational status than father slide into poverty in South Asia. This is likely to .8 be the case for Nepal as well, although a multi- year study the World Bank has recently initiated will generate richer information on the range of vulnerabilities Nepalis face. Those escaping .6 poverty constitute a ballooning vulnerable class. Predicted probability Considering households with a larger than 10 percent probability of falling back into poverty .4 as “vulnerable” households, we see that as the proportion of the poor has declined over the years, there has been a corresponding growth in the size of the vulnerable population from 28 .2 percent in 1995-96 to 45 percent in 2010-11. During the same period, there has also been 0 quite a remarkable growth in the size of the mid- Hill Brah. Hill Chhe. Terai B&MC Dalit Newar Hill Jana. Terai Jana. Musl.& Oth. dle class population in Nepal with the propor- tion increasing from 7 percent in 1995-96 to 22 Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and percent in 2010-11. Close to half of the urban Economic Mobility in Nepal, 2014 Notes: Caste/Ethnicity categories are hill Brahmin, hill Chhettri, Terai Brahmin and Middle population is in the middle class while half of Castes, Dalit, Newar, Hill Janajati, Terai Janajati, Muslim and Others. rural areas is in the vulnerable category. Middle class prevalence is highest in the central region while the mid and far western regions have the highest incidence of poverty and vulnerability. Middle class households are significantly more likely to be headed by those with more than 11 years of education, engaged in non-agricultural employment and more likely to be in white collar jobs. Vulnerability on the other hand is highest among those self-employed in agricul- ture and service sector employment, most likely in blue collar occupations. Overall, poverty and 9 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L vulnerability is highest among households deriv- expected position five year hence. Based on re- ing their livelihoods from wage employment in sponses to these three questions, Gallup defines agriculture. household as thriving, struggling or suffering. Thriving households have strong current life Results from perceptions surveys such as Gal- situation and have positive views about the next lup suggest that this growing middle class is 5 years. Struggling households have wellbeing fundamentally insecure about the prospects that is moderate but inconsistent while suffering of continued improvements in its living stan- households have well-being that is at high risk, dards. Gallup World Survey asks respondents they report poor ratings of their current life and questions about their perceived position in an also poor outlook for the future. Based on this 11-step ladder denoting various parts of the data, 90 percent of Nepalis are either suffering or economic distribution within each country. Re- struggling. This result suggests that even though spondents are asked about current position as there is a nascent middle class building up in well as positions five year before the survey and Nepal, at least half of this middle class is funda- mentally insecure about the prospect of contin- ued improvements in living standards. FIGURE 0.7: EPISODES OF MOVEMENTS IN AND OUT OF POVERTY (% OF POPULATION) In the qualitative survey carried out for this (a) Episode 1995/96-2003/04 analysis, respondents were asked about what they saw as key characteristics of the middle class in Nepal. In in-depth interviews and fo- cus group discussions, an overwhelming major- ity of respondents/participants associated be- ing in the middle class with being engaged in relatively secure professions such as government jobs, teaching, medical and engineering profes- sions, and other forms of private sector salaried employment. This association was reinforced by widespread recognition among respondents that economic prospects of those in this group hinges more critically on professional skills and labor as opposed to things like inherited capital and other forms of financial assets. (b) Episode 2003/04-2010/11 The Productivity Constraint One of the main findings of this report is that liv- ing standards have improved but not sufficiently enough to move those who have escaped pov- erty to a more secure middle class. As a result, a majority of the population today constitutes a vulnerable group that is either perilously close to falling back into poverty or otherwise struggling to cement their economic security. Building prosperity for a majority of Nepalis will entail not just safeguarding the gains already made, but also boosting productivity of the economy in a manner that will help them realize higher income levels. The direct and indirect effects of work related migration, primarily to overseas destinations, has been a critical driver of mobil- Source: World Bank staff calculations using data from three rounds of NLSS data for 1995/96, ity experienced by Nepalis over their lifetimes 2003/04 and 2010/11. The methodology used is called the synthetic panel methodology and across generations. But one key limitation 10 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L of the welfare improvements that have been re- ed a much needed safety valve for the economy. alized has been that these private gains have not At the micro level, the flow of remittances has di- translated into an aggregate improvement in the rectly benefited migrant households. Even house- productive potential of the economy in a man- holds without migrants have benefited from the ner that would generate good jobs for those that growth in labor income. On the one hand, wag- remain in Nepal. es in agriculture have gone up as a result of the tightening of labor supply. On the other hand, Nepal’s atypical structural transformation has increase in the demand for non-agricultural been one major impediment to the development goods and services has led to an increase in de- of the middle class. Agriculture, which used to mand for non-farm labor and pushed up non- command a dominant share of the total value farm wages. In the urban and newly urbanizing added in the economy has declined to 34 per- areas, remittances have been the critical source cent. The share of industries rose and reached of liquidity fueling growth in banking and finan- a peak of 23 percent around the beginning of cial services, restaurants and trade as well as real the Maoist conflict in 1996, but has dwindled estate and construction. The latter in particu- to a low 16 percent since then. In the context of lar, has generated tremendous growth in rents shrinking agriculture and timid industries, ser- to the owners of land and capital within cities vices have grown in prominence and account for transmitting a multiplier effect on the demand more than half of the GDP today. The typical for goods and services. This has aggregated up story of structural transformation is one through to the macro level where the remittance receipts which workers leaving low productivity agri- have been critical to maintaining Nepal’s cur- culture in favor of higher productivity jobs in rent account position, despite a widening trade manufacturing agglomerate in cities which act deficit as the country imports a growing amount as production centers and engines for growth. of consumer goods ranging from the most basic In the case of Nepal, the economy has diversified such as rice and petroleum products to telecom- away from agriculture over the long run, but it munications equipment, luxury cars and parts. appears to have skipped the intermediate manu- The overall surplus position in the balance of facturing phase and gone directly to services. payments, largely due to remittance inflows have This “premature graduation” from industries been a critical buffer against global macroeco- towards low skilled services is one potential rea- nomic shocks. son for the apparent disconnect between private income growth and the formation of a robust FIGURE 0.8: THE POOR, VULNERABLE AND THE MIDDLE CLASS IN NEPAL middle class. 100% 7 Despite the dramatic shift in the production 90% 14 22 base of the economy, the change in employ- 80% 28 ment patterns have been relatively modest. A 70% 35 sizeable majority of rural Nepalis continue to 60% be self-employed in agriculture. The most sig- 45 50% nificant change in the rural areas is the increase 40% in off-farm self-employment which has roughly 64 30% doubled from 4.5 percent in 1995 to 8.1 per- 50 20% cent in 2010. Focusing only on wage jobs, the 31 10% most dramatic increase in wage employment in rural areas has come in construction which 0% 1995/962 003/04 2010/11 is mostly of a casual nature. White collar jobs Poor Vulnerable Middle Class Upper in professional services and manufacturing have increased in urban areas but the overall skill con- Source: World Bank staff estimates based on data from three rounds of NLSS data for tent of all wage jobs has been stagnant over the 1995/96, 2003/04 and 2010/11 Note: The consumption aggregate used for this part of the analysis is the 30-day recall version period. which is not the version used for official poverty estimates for 2010/11. Given the low level of job creation in the econo- my, overseas work and remittances have provid- 11 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L But in a perverse way, remittances have also development potential. A stable, positive bal- weighed upon Nepal’s development potential. ance of payment position and a strong fiscal po- First, the buoyant influence remittances have sition maintained in part by the revenues from had on the purchasing power of a majority of taxes on an ever growing base of imports have Nepalis has inadvertently also resulted in the been the bulwarks of Nepal’s overall macroeco- appreciation of the country’s real exchange rate nomic stability over the last decade. which has in turn eroded the competitive edge of manufacturing industries, further undermin- But remittance driven growth is not likely to ing the employment generation potential of this continue forever, especially given the already sector. Second, the steady outflow of young and high base. Further, there are signs that the rate potentially restless youth and the cushion of re- of migrant outflow is beginning to taper off as mittances could have also lulled policymakers the major host countries show signs of stress in into not seeing the immediate urgency of key the face of declining crude oil prices. With slow- reforms necessary to unlock Nepal’s longer term down in migration, a slowdown in the growth FIGURE 0.9: CHARACTERISTICS OF THE POOR, VULNERABLE AND THE MIDDLE CLASS, 2011 (a) By geographic region (b) By education and occupation 0% 10%2 0% 30%4 0% 50%6 0% 70%8 0% 90%1 00% Agri Self 35 49 15 Servi. Self 21 30 42 Agri Employ 58 38 3 Professional Employ 8 31 54 Other services Employ 33 46 20 0 No schooling 42 46 12 11+ years of schooling 7 29 57 Poor Vulnerable Middle class Upper class Source: World Bank staff estimates based on data from three rounds of NLSS data for 1995/96, 2003/04 and 2010/11 12 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L of remittances is inevitable. While the strategy As the primary seat of the government, Kath- Greater has – by default – served as a useful buffer dur- mandu has enjoyed historical primacy among rural-urban ing turbulent times for Nepal, it cannot be the cities but over the last two decades, its share of durable source of growth required to turn Nepal total urban population has been declining. In migration and into a middle income country with a vibrant 1991, Kathmandu accounted for 25 percent of the emergence middle class by 2030. Nepal’s urban population. By 2011, this num- urban sprawls ber was down to 22 percent. Considering Kath- around main Urban agglomeration – A mandu valley as a whole, i.e., municipalities of Missed Opportunity? Kathmandu, Lalitpur, Bhaktapur, Kirtipur and highways Urbanization has increased quite rapidly. In Madhyapur Thimi taken as a single urban clus- bordering India fact, Nepal is the fastest urbanizing country in ter, the decline is even sharper and goes from and district South Asia. Greater rural-urban migration and 39 percent in 1991 to 32 percent in 2011. Cit- headquarters the emergence of urban sprawls around main ies play an important role in driving growth. highways bordering India and district head- For Nepal, the emergence of unplanned urban have been quarters have been the key characteristics of clusters and secondary towns raises the question the key urban growth. One crucial driver of the recent of whether the loss of primacy of the country’s characteristics of urbanization trends has been the growth in rural pre-eminent urban cluster is an entirely desir- urban growth. incomes and the transmittal of tastes and prefer- able outcome from the point of view of growth. ences for certain amenities including basic ser- vices such as better education and health that Directions for Policy are more likely to be serviced in urban clusters. So what does Nepal need to do to consolidate While rural-urban migration, together with mi- the gains on improving living standards, reduc- gration to India, still represents an important ing poverty and boosting shared prosperity? income diversification strategy for households Three main policy directions emerge from this not directly able to participate in international analysis. migration due to cost constraints, most of these domestic migrants find employment in low First, equalizing opportunities and leveling the skilled services such as construction, trade and playing field to enable Nepalis from all walks of retail and hospitality sectors in the cities. life to participate to economic life is not only a sensible way to move forward on the inclu- The pull of a dynamic urban economy generat- sion agenda, but it is also likely to be good for ing good jobs appears to be largely non-existent growth. All children, including Champa and and the dominance of push-factors in driving Goma must have as much as a decent shot at urban growth implies that the consumer ad- vantage of cities and newly emerging towns has FIGURE 0.10: THE PROPORTION OF NEPALIS STRUGGLING, SUFFERING AND THRIVING not translated into a productive advantage. One evidence on this is the spatial pattern of wage growth. Wages have grown across the country, 0.8 but the pace of growth has been faster in rural 0.7 0.7 areas and the newer towns than in core metro- 0.6 politan areas of Kathmandu and other urban centers. Taking into account the cost of living 0.5 differences across the country, the pace of wage 0.4 growth in Kathmandu in particular, relative to the rest of the country has been miniscule. The 0.3 0.2 negative effects of agglomeration, on the other 0.2 hand, have been highly visible as the unplanned 0.1 and unregulated growth of the city has not only 0.1 led to the deterioration of basic services such as 0.0 drinking water and sanitation but also polluted Struggling Thriving the city’s rivers and air to highly toxic levels. Source: World Bank staff estimates based on data from Gallup World Survey. 13 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Nepal’s realizing their life’s potential as Avidit. This will there are already provisions for affirmative ac- require a three pronged effort. First, improving tion in civil service appointments. But ensuring decentralization the quality of service delivery generally remains adherence to these principles in legislation and efforts have a critical part of the agenda on improving op- practice would go a long way. been stunted portunities. Nepal’s decentralization efforts have for a long time been stunted for a long time due to conflict and To reduce vulnerabilities Nepal needs to political turmoil. But with the country having strengthen safety nets systems, particularly those due to conflict adopted a federal structure in its new constitu- that insure households against a variety of in- and political tion, strengthening local accountability through come shocks. There is a high degree of vulner- turmoil. the electoral process and building in the inequal- ability and transitions around the poverty line ity of opportunities into formulas that will guide with one person falling back into poverty for intergovernmental fiscal transfers will be two pos- every two that escape. Given the experience of sible ways of improving services, particularly in the recent earthquake, one area that appears par- the underserved regions. Second, given inequali- ticularly important is the design of systems that ties within regions, there is also an urgent need will enable the government to respond in a more to strengthen equalizing investments, through agile and efficient manner, not just to catastroph- scholarships, health insurance and social safety ic disasters of similar scale but also to smaller but nets that should ideally be targeted on the basis more frequent shocks, or the dull disasters. on economic need. Finally, in order to redress the residual effects of past disadvantages, some Designing safety net systems that are able to form of affirmative action may have to be main- target and deliver assistance based on broader, tained. The new constitution has made provi- income based notions of poverty and vulner- sions for representation of underserved groups ability is also an important agenda because the at various levels of political and public office and existing programs have had very limited impact FIGURE 0.11: CHARACTERISTICS OF THE POOR, VULNERABLE AND THE MIDDLE CLASS, 2011 Earthquake 60000 50000 Monthly outflow of migrant workers 40000 30000 20000 10000 January/08 January/10 January/12 January/14 January/16 Source: Ministry of Foreign Employment Database Note: The solid line represents the trend obtained by applying the Hodrick-Prescott filter to the raw data. 14 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L on poverty. Investments in these activities may strategy. The lack of adequate number of good appear to be a drain on the budget and indeed jobs appears to be the fundamental constraint the issue of fiscal space is an important consid- to the formation of a robust middle class in Ne- eration. But two mitigating arguments need to pal. To that end stabilizing the political environ- be considered. First, significant savings may be ment, carrying out the necessary reforms to im- possible by rationalizing the highly fragmented prove the investment climate and addressing the current portfolio of social assistance. Second, crippling infrastructure deficit would be equally given the scale of vulnerabilities Nepali house- essential entry points. holds routinely deal with, helping protect their savings and assets during times of crisis, could Leveraging the agglomeration potential of Ne- The lack of potentially enhance their ability to take risks pal’s cities would also be an important policy adequate and make productivity enhancing investments direction. The current industrial policy provides in both agriculture and non-agricultural activi- tax incentives and subsidy for investment in the number of good ties which could actually help growth. development of the plant for firms to locate their jobs is the main productive activities in lagging regions. These constraint to the Finally, on enhancing productivity, three key “spatially targeted” policies represent the gov- formation of a policy directions are identified. First, as most ernment’s deliberate efforts to de-concentrate of Nepal’s chronic poor remain in agriculture, productive activities. But locating production robust middle there is an urgent need to boost productivity activities in remote locations with lower popula- class in Nepal. in this sector. This will require, among others, tion densities is seldom effective in propping up promoting diversification from cereal crops to lagging regions or generating growth. In fact, if cash crops, greater commercialization and bet- enforced, such policies could be detrimental for ter integration with markets and value chains. the growth of the industrial sector. Instead, from As the overseas migration experience deepens, the perspective of growth, a better policy would the pool of migrant returnees is also likely to in- be to invest in enhancing connectivity of the crease over time. These migrants will come not remote regions to the emerging growth centers only with resources but also with creative ideas or to invest in the development of these growth and visions formulated through rich experiences centers themselves. working abroad. Having the right policy and in- vestment environment as well as a whole host of Finally, it must also be recognized that there is no complementary inputs such as better connectiv- level of growth Nepal can realistically generate in ity through rural roads and financial services will the medium term that will generate jobs at a fast help leverage their creativity and entrepreneurial enough pace to completely overturn the flow of spirit to boost growth in this sector. migrant workers. A more realistic outlook is that some level of overseas migration will remain a Second, in order to accelerate the creation of critical dimension of Nepal’s jobs environment in productive jobs, Nepal must figure out a way the medium to long run. For this reason, building to generate more and better jobs within Nepal. a better understanding of not just the migration Whether this is going to be by putting in place process in general but also of what can be done to policies to rekindle growth, particularly in its make the process more efficient and less costly so industrial/manufacturing sectors or through that successive cohorts of Nepali migrants have exploring avenues for promoting higher value better information of opportunities abroad, can services that have some degree of tradability become more productive and can have access to or a combination of the two is something that better quality jobs abroad is also an important should come out in a clearly articulated growth part of the jobs agenda. 15 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 16 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L I. Poverty and the Drivers of Poverty Reduction Turbulent politics, weak growth timate puts economic growth for fiscal year The year 2015 was rough on Nepal. On April 2014/15 at 0.77, a fourteen year low. 25, a catastrophic earthquake of magnitude 7.8 struck the country. This was followed by a series As devastating as these recent shocks have been, of aftershocks including a particularly large one, in a way, they are dramatic manifestations of measuring 7.3 magnitude on May 12. A total how conflict and fragility have more generally of 8,790 people lost their lives and more than impinged on Nepal’s development potential 22,300 suffered injuries. An estimated 500,000 over the last two decades. More than 14,000 dwellings were completely destroyed and more Nepalis lost their lives between 1996 and 2006 than 8 million people (roughly a third of the in a violent Maoist conflict that plunged the country’s population) were directly affected. As country into a civil war. Even though that crisis the country pulled itself together in the after- was averted, the ensuing years of mainstream- math, the shifting political landscape provided ing the Maoists into the political process and a rare opening in the constitutional process that drafting a new constitution were characterized had been in abeyance for the past eight years. The by two rounds of elections to the constituent major political parties moved speedily to finalize assembly, frequent changes in political alliances and adopt the new constitution, but clashes broke and government priorities and public discourse out in the southern belt of the country where dominated by matters of politics as opposed to large segments of the population were unhappy economics. The lack of stability in the political with certain provisions in the new constitution. sphere took a toll on the already anemic invest- The constitution was eventually adopted in Sep- tember, but the resulting protests and disruptions FIGURE 1: WHERE WOULD NEPAL BE IF IT HAD GROWN AT THE brought Nepal’s cross-border trade with India to RATE AT WHICH SOME NEIGHBORS HAVE GROWN? a complete standstill for a period of four months. This was bittersweet for a majority of Nepal is eagerly awaiting a resolution to what has been a rather prolonged period of political transition. In a single generation, the country has gone from being a unitary, Hindu state ruled by an absolute monarch to a secular republic that has adopted a federal structure. The promulgation of the constitution was supposed to launch the country back on a path towards stability and prosperity. Instead it was off to a shaky start, with the trade disruption severely affecting eco- nomic activities across the board. Fuel and other essential commodities were acutely in short sup- ply. Shortage of raw materials and essential fuel forced businesses to either shut down entirely or operate at reduced capacity. Owing largely to these disruptions, the latest government es- Source: WB Staff calculations using WDI, 2016 17 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L ment climate in the country. As a result, the ever, the dramatic changes in consumption pat- economy sputtered in fits and starts around a terns of Nepalis observed in 2010-11 prompted mean per capita growth rate of 2.5 percent be- the government to revise the poverty line. With tween 2005 and 2015. the revised poverty line that effectively increased the real value of the line by 35 percent from the Decline in This is low in comparison to progress made by previous line, consumption poverty in Nepal other South Asian neighbors in recent years. stood at 25.2 percent in 2010-11.2 Regardless poverty has been At the onset of the civil war in 1995, the dif- of which poverty line is used, the rate of pov- accompanied by ference in GDP per capita expressed in 2010 erty reduction between 2003-04 and 2010-11 improvements USD between Nepal and Bangladesh was $44. was almost twice the rate observed in the previ- in durable asset A decade later, in 2005, this gap had widened ous period and over the 15 year period between to almost $100. By 2015, the difference in per 1995 and 2010, absolute poverty declined ownership. capita GDP stood at $283. The growth rate that steadily by around 2.2 percentage points a year. Nepal’s northern neighbor, China, has been able (Table 1) The rate of poverty decline seen in Ne- to generate and sustain is no doubt an unreal- pal, particularly during the second episode, was istic target. But if Nepal had simply been able among the fastest in the world. One global pov- to emulate the growth rates seen in India and erty reduction benchmarking exercise puts Ne- Cambodia (a country that emerged from a pal in the 96th percentile of all available country deeper conflict in the early 1990s), its per capita episodes.3 GDP today would be over $1000. (Figure 1) A recent World Bank assessment finds that the This general improvement in living standards is current growth trajectory will keep Nepal a low corroborated by trends in ownership of private income country until 2025.1 assets among the different income groups over 15 year period. Everyday items such as kitchen Income poverty has declined utensils are universally owned in all three years A surprising contrast to this somewhat grim while the ownership of cars and washing ma- narrative on missed opportunities on econom- chines is rare. For most other commodities, it ic growth has been the stellar progress Nepal appears that there is sharp increase in ownership has made in the social sectors, particularly on first among the top 20 percent between 1995- poverty and human development. Poverty has 96 and 2003-04 and followed by the middle 40 continued to decline, at a faster rate in recent percent and then the bottom 40 percent in lat- years than in the past. Between 1995-96 and ter period. In 1995-96, few households in the 2003-04, consumption based poverty declined bottom 40 percent and the middle 40 percent from 41.8 percent to 30.8 percent. This decline owned items such as phones, TV/VCR/VCD continued in 2010-11 – the most recent year for and fans but their ownership increased drasti- which data on household surveys exists – and cally among the middle 40 percent by 2010-11. would have reached a low of 12.5 percent using On the other hand, ownership of more expen- the poverty line established in 1995-96. How- sive items, such as motorcycle/scooter, refrig- TABLE 1. POVERTY RATES AND PACE OF POVERTY REDUCTION IN NEPAL 1995/96 2010/11 Official Poverty Survey Year Poverty Line Poverty Line Estimate in 2010/11 Poverty Headcount Rate (% of Population) 1995-96 41.8% 63.8%   2003-04 30.8% 49.4%   2010-11 12.5% 30.8% 25.2% Period Pace of Poverty Reduction (annual percentage point change) 95/96 - 03/04 1.36% 1.80%   03/04 - 10/11 2.62% 2.66%   95/96 – 10/11 1.96% 2.20% Source: World Bank (2013) 18 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L erator/freezer and computer improved but only Index (MPI).4 (Figure 3) Significant improve- exclusively among the top 20 percent. (Figure 2) ments have been made in dimensions such as education, health and access to basic services. … so has non-income based For example, households with school-aged chil- measure of poverty dren (6 to 15 years old) out of school declined The improvements in living standards along from 55 percent in 1995-96 to 16 percent in monetary measures of welfare have been ac- 2010-11. This is presumably due to improve- companied by improvements in non-income di- ments in accessibility to and availability of mensions of well-being that would typically be school, as households with children spending used to construct a Multidimensional Poverty more than 20 minutes to reach primary school FIGURE 2. ASSET OWNERSHIP AMONG THE VARIOUS INCOME GROUPS Source: World Bank Staff Calculation based on Nepal Living Standard Surveys 1995/96, 2003/04, and 2010/11. Question on computer ownership was not available in 1995/96. 2014/15 estimates are added when data is available in the 2014/15 Annual Household Survey. 19 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 3. REDUCED DEPRIVATIONS IN NON-INCOME DIMENSIONS BETWEEN 1995-96 AND 2010-11 100 90 80 70 60 50 40 30 20 10 0 Child out of schoolT ime to primary Undernourishment No electricity No phones at school for delivery cooking fuel home 1995-96 2010-11 Source: World Bank (2013) declined from 84 percent to 33 percent during 2011, every fourth household in Nepal had an the same period. But much of it is also likely overseas migrant and every fifth household had a due to increased ability to afford schools, includ- migrant in a country other than India.5 Increase ing private schools. In 1995-96, almost one in in overseas migration to destinations other than every two households experienced a child birth India has in fact been the key driver of the recent without the presence of a medical professional. increase in migration in Nepal. This declined to only one in five by 2010-11. Similarly, access to sanitation, electricity, and This migrant population is predominantly telecommunications services also improved ap- young and male. In 2011, about 90 percent of preciably. For example, approximately four out the migrants outside India were male. The me- of five households did not have an in-home toi- dian migrant is only 25 years old and almost 90 let in 1995-96. By 2010-11, two out of three percent of migrants are aged between 15 and 45 households had one. Telecommunications ser- years. Migrant males account for 13 percent of vices – measured by having a telephone in the the resident male population. Since migrants home – used to be a preserve of the top 1 per- are mostly of working age, more than a fifth (22 cent in 1995-96. By 2010-11, 74 percent of Ne- percent) of the national (resident and migrant) palis had tele-connectivity in their homes. population aged 15-45 is outside the country. The educated migrants, especially outside India, Remittances have played a part earn more and remit more money to their fami- The most commonly cited driver of the observed lies in Nepal. Most of the migrants migrated poverty reduction is the dramatic increase in re- for work. For example, over 97 percent of the mittances received from abroad since the late migrants to Malaysia and the Gulf countries mi- 1990’s (World Bank, 2006, 2013). This is the grated for work. result of a steady increase in overseas migration over the past decade and half. Between 1981 With the increase in migration, remittances and 2001, only about 3 percent of the popu- from abroad have also increased. Until the late lation was abroad, with a cumulative increase 1990’s, personal remittances were less than one of 0.7 percentage points in 20 years. By 2011, percent of GDP, lower than Bangladesh or In- over 7 percent of the population was abroad. In dia. The first half of the 2000’s saw a drastic 20 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 4. PERSONAL REMITTANCES RECEIVED IN NEPAL & SELECT COUNTRIES Source: World Development Indicators (World Bank, 2016), using personal remittances received as % of GDP (BX.TRF.PWKR.DT.GD.ZS), personal remittances received in current US dollars (BX.TRF.PWKR.CD.DT) and the total population (SP.POP.TOTL). increase in this share, from 2 percent in 2000 an important part in improving the living stan- to 15 percent in 2005, 22 percent in 2010 dards witnessed in Nepal. and as much as 29 percent in 2014, while the neighboring countries experienced an increase So have opportunities within at a much more modest rate (Figure 4).6 In per Nepal for non-migrants capita terms, Nepal received less than $5 of re- Households within Nepal have obviously ben- mittances from abroad in 2000. That number efited from private receipts of remittances. But had reached $205 by 2014, more than twice even non-migrant households have benefited as much as Bangladesh (94 dollars) which has from opportunities created by the changes in the second highest remittance receiving country the structure of the domestic economy, brought in South Asia. This figure itself may be heavily about, indirectly, by migration and remittances. understated because official numbers on remit- For example, the large exodus of young, produc- tance flows through the banking system do not tive and predominantly male population could capture informal flows that are estimated to be affect labor supply and wages, particularly in quite large as well. agricultural sectors. Similarly, the remittance led growth in income and consumption in the rela- Unlike other growth processes, the benefits of the tively isolated, largely unconnected local econo- growth in remittances have directly accrued to mies could fundamentally alter the landscape households. In 1995-96, approximately one in for the demand for agricultural as well as non- four households received some form of remit- agricultural commodities and also the demand tances. This became one in three by 2003-04 for labor employed in these sectors. The result- and more than one in two by 2010-11. House- ing changes in wages and prices of other factors holds across the consumption distribution have as well as other outputs could have ended up benefited from the growth in remittances. (Fig- benefiting non-migrant households indirectly. ure 5a) Not only did more Nepali households start receiving remittances, the amount of remit- Indeed, decomposition exercises show that tances received by remittance receiving house- half of the poverty reduction in Nepal between holds also increased over the years (Figure 5b). 1995-96 and 2010-11 was be driven by an in- Given the size and nature of this increase, it is crease in labor income which includes income entirely plausible for remittances to have played from wages as well as self-employment. This is 21 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 5. INCIDENCE OF REMITTANCE ACROSS THE CONSUMPTION DISTRIBUTION (a) More households receive remittance NLSS 1995/96 NLSS 2003/04 NLSS 2010/11 (b) More remittances are received (1995-2010) 20 Annual per-capita growth (%) 15 10 5 0 20 40 60 80 100 Consumption Percentiles Source: World Development Indicators (World Bank, 2016), using personal remittances received as % of GDP (BX.TRF.PWKR.DT.GD.ZS), personal remittances received in current US dollars (BX.TRF.PWKR.CD.DT) and the total population (SP.POP.TOTL). 22 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L particularly the case in rural areas. Remittances FIGURE 6. DRIVERS OF POVERTY REDUCTION BETWEEN 1995-96 AND 2010-11 are important. But directly they account for only (a) By labor and other sources of income a fourth of the poverty reduction observed na- tionally and about a third in rural areas. (Figure 6a)The relatively small contribution of remit- tances may come as a surprise but given the high up-front costs involved, the poorest segments of the population are least likely to be able to ben- efit from the newer, non-India migration desti- nations that have been the drivers of the growth in remittance income at the household level. Further, breaking down the sources of income between farm and non-farm income, non-farm income growth appears to have been more im- portant for the observed reduction in poverty than farm incomes, though farm incomes have contributed quite a bit as well. (Figure 6b)This is consistent with the observed diversification of the rural economies with increased non-farm (b) By agricultural and non-agricultural sources of income activity during the period. Changes in the de- mographic structure of the household – which captures, among others, the dependency struc- ture or the number of household members that participate in the labor market relative to the number of dependents they have – also appears to have been important for poverty reduction. The primary reason for the dominant role of labor income in the observed poverty reduc- tion has been the increase in wages across the country. (Figure 7) Mean real wages grew from Rs. 184 per day to Rs. 326 per day (in 2010 Rupees) or roughly 5.1 percent per year over this period. Agricultural wages (4.2 percent) grew faster than the non-agricultural wages (3.3 percent). Increased overseas migration of work- Source: World Bank staff calculations based on data from NLSS 1995-96 and 2010-11 Note: Capital income and other transfers arein the residual. ing age males and the resulting tightening of the supply of labor, particularly in the farm sector are possible drivers of the increase in agricultural Another important but often under-appreciated wages. The increase in non-agricultural wages driver of Nepal’s poverty reduction in the last on the other hand are largely due to the growth two decades has been the country’s demographic in the demand for non-farm labor which was in transition. Nepal’s population growth peaked turn driven by economic diversification and in- around 2.7 percent in the early 1990s and was at crease in non-farm activities brought about by sub-Saharan country average at the time. Since remittance induced growth in incomes and the then it has taken a steep downward turn and set- associated changes in demand for goods and ser- tled at around the global average of 1.2 percent a vices. Non-wage employment in off farm sectors year in the late 2000s. The decline in population has also increased suggesting that opportunities growth rate has coincided with a sharp decline created outside of agriculture, particularly in ru- in fertility rates. From over 6 until the 1970s ral areas has been the key ticket out of poverty in and over 5 until the 1990s, the number of births the preceding two decades. of an average Nepali woman had declined to 23 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 24 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L around 2 in 2014. Demographers have various- FIGURE 7. MEAN DAILY WAGES IN AGRICULTURE AND NON-AGRICULTURE ly attributed this fertility transition observed in Nepal to the spread of mass education, increas- ing female labor participation, mechanization of agriculture, among others. The increase in urbanization is also another po- tential driver. The share of urban population has been increasing steadily, from 3 percent in the 1960s, 9 percent by 1990, and 18 percent by 2014. This is the fastest urban growth rate in all of South Asia.7 (Muzzini & Aparicio, 2013) More recently, the social and economic changes brought about by the increase in the migration of overwhelmingly male population together with an increase in educational attainment of girls has increased the average age of first marriage. Whatever the reasons may be, the fertility tran- sition seen in Nepal in the last two decades has been quite dramatic and the reduction in aver- Source: World Bank staff calculations using data from NLSS 1995-96 and 2010-11. age household sizes, especially among the poorer, Note: A rough rule of thumb to relate to these numbers in terms of 2015 Rupees is to multiply them by 1.5 which is the factor by which CPI has grown during this period. rural population, together with the declining dependency ratio has also contributed to the ob- served poverty reduction quite appreciably. FIGURE 8. POPULATION AND POPULATION GROWTH IN NEPAL 30 7 Population Growth (%)/Fertility Rate (birth per woman) 6 25 5 20 Total Population (millions) 4 15 3 10 2 5 1 0 0 Source: World Bank staff calculations using data from NLSS 1995-96 and 2010-11. Note: A rough rule of thumb to relate to these numbers in terms of 2015 Rupees is to multiply them by 1.5 which is the factor by which CPI has grown during this period. 25 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 26 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L II. INEQUALITY AND THE DEMAND FOR INCLUSION Demand for inclusion Exclusion – or the lack of inclusion – is no doubt dominates political and a broader concept than inequality. Importantly, economic discourse it incorporates notions of dignity in the partici- A question that follows is the extent to which pation in social and economic life, something a the progress on living standards improvements purely economic measure of inequality does not has been realized evenly by all segments of the capture. Yet, given Nepal’s difficult geography population. This question has a special signifi- and a legacy of exclusion along the dimensions cance for a country that has a violent Maoist of caste and ethnicity, the concepts overlap quite conflict in its recent history. Many have postu- a bit, at least in the way they are commonly per- lated the perceived inequality in the distribution ceived. So to what extent is the increased aware- of resources as being one of key initial drivers ness about exclusion or the increased demand of the revolution that turned into a civil war for inclusion correlated with the level and trends resulting in 14,000 deaths between 1996 and of inequality in Nepal? How high is inequality 2006. And indeed, if there is one agenda that in Nepal and has inequality increased? has dominated Nepal’s political and economic discourse during the decade of post-conflict Conventional measures such as the Gini index transition, a period that has seen the dissolution suggest that overall inequality in Nepal has been of the 250-year old institution of monarchy and low and stable over the 15 year period between a deep restructuring of the Nepali state, it is the 1995-96 and 2010-11. The Gini index for con- agenda of inclusion. sumption for Nepal was 0.33 in 2010-11, which is at the lower end of global range for this num- The salience of the agenda is reflected quite no- ber. This measure of inequality also remained tably in the recently promulgated 2015 Con- virtually unchanged between 1995-96 and stitution of Nepal. The word “inclusion” or its variant “inclusive” appears in 35 instances in FIGURE 9. FREQUENCY OF USE OF SOME KEY WORDS IN THE 1990 AND the translated document. These words did not 2015 CONSTITUTIONS OF NEPAL, USE PER 1000 WORDS appear at all in the earlier democratic constitu- tion of 1990. Likewise, the word “equal” and its variants “equality” and “equity” are used 38 times (compared to 8 times in 1990) and the word “rights” is used 107 times (compared to 11 times in 1990). Now, the 2015 Constitution of Nepal is admittedly a longer document; it con- tains roughly twice as many words as the ear- lier Constitution. But even after accounting for that, words denoting related concepts of equity, equality, rights and inclusion appear to have been used with a significantly higher frequency than in the past. In contrast, the use of words such as “justice” and “sovereignty”, which are often common words in constitutions, roughly Source: World Bank staff calculations based on unofficial translations available on the internet. stayed the same or declined. 27 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 2010-11 both at the national level and in rural TABLE 2. GINI INDEX OF CONSUMPTION IN areas.8 In the urban areas, there appears to have NEPAL IN 1995-96 AND 2010-11 been a sharp decrease. (Table 2) Consumption Aggregate Old New Measured this way, inequality in Nepal is low Survey Year 1995/96 2010/11 2010/11 not only in absolute terms, it is low also relative to the level of its economic development. This National 35% 35% 33% is an inequality phenomenon that Nepal shares Urban 46% 36% 35% with other South Asian countries as well. Ne- Rural 31% 31% 31% pal’s Gini index is one of the lowest among the Source: Nepal Living Standard Survey in 1995/96 and 2010/11 countries with a similar level of GNI per capita, Note: The difference between the “Old” and “New” consumption most of which are countries in Sub-Saharan Af- aggregate is that the new one uses a 7-day recall for food expenditures. rica, and well within the range of more advanced countries in Europe and Central Asia. If one is richest 10 percent of the population often holds to imagine a cross-country Kuznets curve over more than 40 percent of the total wealth and this plot, it would appear that Nepal is at the the poorest 10 percent explains two percent or very early stages of development and that the less. Many countries in Sub-Saharan Africa and engine of growth has not started cranking up to Latin America and Caribbean regions fall under the extent where this growth acceleration would this category. In the case of Nepal, the consump- be begin to generate a widening consumption tion share held by the top 10 percent is one of distribution. the lowest at 26 percent and the share by the bottom 10 percent is one of the highest at 3.5 Other standard measures of inequality also sug- percent, another indication of low overall in- gest low and stable inequality in Nepal. The equality. (Figure 11) shares of welfare held by the richest and the poorest groups in the country represent in- The natural question to ask next is whether the equality at the tail ends of the distribution. In inequality in income mirrors what we conclude countries plagued with extreme inequality, the about inequality of consumption. Perceptions FIGURE 10. GINI INDEX BY THE LEVEL OF DEVELOPMENT East Asia & Pacific Europe & Central Asia 60 Latin America & Caribbean Middle East & North Africa North America 50 South Asia Sub-Saharan Africa 40 Gini Index 30 20 Nepal (2010) Gini: 32.75 GNI per capita (Atlas Method): 540 10 10 100 1,000 10,000 100,000 GNI per capita (Atlas Method) on Log Scale Source: World Development Indicators (World Bank, 2016), based on the latest data for all available countries. 28 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 11. WELFARE SHARES BY THE TOP AND BOTTOM 10 PERCENT Welfare Share Held by Top 10 Percent Welfare Share Held by Bottom 10 Percent Pakistan (2010) Bangladesh (2010) India (2011) Nepal (2010) Share held by Bottom 10 percent: 3.5 Pakistan (2010) Nepal (2010) Share held by Top 10 percent: 26 Sri Lanka (2012) Bangladesh (2010) Bhutan (2012) Maldives (2009) Maldives (2009) India (2011) Bhutan (2012) Sri Lanka (2012) East Asia & Pacific Europe & Central Asia Latin America & Caribbean Middle East & North Africa North America South Asia Sub-Saharan Africa 0 10 20 30 40 50 60 70 0 12 3 4 5 6 7 89 Source: World Development Indicators (World Bank, 2016), using all available countries for which the latest data is 2000 or later about inequality are often more closely aligned ity (Deaton and Zaidi, 2002; Haughton and with income inequality than the inequality of Khandker, 2009; World Bank 2015) because it consumption or living standards. Fortunately fluctuates much less and is believed to be a bet- for Nepal, the past three rounds of the Nepal ter approximation of the concept of permanent Living Standards Surveys consistently collected income. Practically, consumption is also less information about respondents’ income from susceptible to measurement errors than income. various sources making an analysis of income High income earners are less likely to respond to inequality possible. It must be noted that inter- income-related questions and tend to understate preting and reconciling differences in inequal- the true income when they do respond. Income ity of income and inequality of consumption is is also notoriously hard to measure for house- often difficult. Theoretically, consumption is the holds dependent on subsistence agriculture and preferred measure for both poverty and inequal- more generally reliant on the informal economy 29 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 12. GROWTH INCIDENCE CURVE OF CONSUMPTION AND INCOME BETWEEN 1995-96 AND 2010-11 Source: World Bank staff calculations based on data from NLSS I and NLSS III making it particularly ill-suited for poverty and forms of affirmative action in higher educa- inequality analysis in a country like Nepal. tion (e.g. medical and engineering colleges) and employment for the disadvantaged groups. Still, with these caveats, the analysis of income Nepal’s newly adopted constitution has made inequality suggests that it may have increased several provisions for affirmative action poli- sharply during this period. (Figure 12) The cies including reservations for underrepresented curve plots growth rates of average income and groups in various levels of political office par- consumption at every percentile and allows for ticularly through the principle of proportional a visual inspection of any conspicuous growth representation. While well intentioned, these patterns across the distribution. Consistent with policies could potentially backfire if they end the abovementioned observations of the stable up de-incentivizing investments and hurting inequality trend, the growth incidence for con- growth. Raising attainment levels of historically sumption is mostly flat across the distribution. disadvantaged social groups is a worthwhile goal This suggests that living standards improved and affirmative action policies could play an im- almost uniformly and at the same rate for all portant role toward that. But in settings where Nepali households. In contrast, the growth in deprivation and disadvantage cut across many income was much faster at the higher end of the dimensions, they might also become difficult to distribution. Income growth appears to have build consensus around without disturbing so- has been slower than consumption growth at cietal harmony. the lower tail of the distribution, but average income has doubled at the 50th percentile and To an extent, this is already playing out in de- tripled at the 95th percentile and higher. bates around several policies that predate the re- cently adopted constitution in Nepal. For exam- Addressing inequality through ple, according to provisions in the Civil Service public policy is difficult Act, 45 percent of new positions are reserved A direct response to inequality is often redistri- for various under-represented groups, women, bution through progressive tax policies in the Dalits, Janajatis, Madhesis, disabled and Nepalis space of income, or through quotas and other living in lagging regions. A segment of the Ne- 30 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L pali population sees this as rightful (if at least But framing the debate along in intention) effort to expand opportunities the lines of equity could for historically disadvantaged groups. Another bridge the divide equally vocal segment of the population objects Alternatively, one could take a step back and ask on the grounds of violation of the principles of a more fundamental question about the kind meritocracy. Yet another point of view acknowl- of inequality that is being measured. This ques- edges that some groups are on average, worse off tion forces us to reorient our discussion from than others, but contends that a sizeable frac- thinking about the extent and magnitude of in- tion of the groups considered privileged are in equality to the underlying type of inequality. In fact very poor and thus should also be covered particular, it is useful to make a distinction be- under the provisions of affirmative action. The tween inequality of outcomes such as income or argument that is often heard is that these caste/ consumption/living standards, and inequality of ethnicity based affirmative action policies end opportunities to arrive at those outcomes. The up willfully neglecting large segments of Brah- focus of development debate, even in Nepal, is mins and Chettris who are economically poor often on inequality of outcomes. But framing and thus deserving of targeted policies but are the same discussion around inequality of oppor- excluded from consideration due to their privi- tunities has an intuitive appeal and could poten- leged caste status.9 tially transcend ideological differences and affect meaningful policy action. The potential for a satisfactory resolution of this debate is further diminished if there are To see why, consider the example of two people disagreements about how well inequalities are who were born with the exact same set of oppor- being measured in the first place. The standard tunities. Most people would probably agree that measures of inequality commonly guiding pub- differences between the two in, say, income, are lic policy come from household surveys such as in part due to differences in life choices, innate the Nepal Living Standards Survey. While these abilities or effort. But now consider two people surveys do a reasonably good job of capturing with identical innate ability but vastly different living standards of those at the low to middle sets of opportunities to begin with. In such a end of the income distribution, they are not very case, it is entirely possible that they end up with effective in capturing the very top end of the in- different outcomes despite exerting the same come/consumption distribution. This is because amount of effort or making similarly sound survey participation and response rates are often life choices. Most people would regard such a very low for wealthier households. This could scenario as unfair. In other words, inequality of cause a downward bias in the measured inequal- opportunities is fundamentally objectionable, ity. A research frontier on this topic explores the whereas inequality of outcomes is objectionable possibility of combining data on income tax re- only to the extent that it is caused by inequal- cords with household surveys to try and get a ity of opportunity. This problem is exacerbated Inequality of handle on what the true distribution of income when the distribution of opportunities is linked opportunities is may look like without the truncated top.10 Un- to circumstances, such as ethnicity, birth, gen- fundamentally fortunately, this is not feasible in Nepal. Sec- der, or family background, because these sys- ond, income is likely to be measured with high objectionable, tematically can impede the upward mobility of measurement errors, especially in countries with an entire group of people. whereas large informal sectors and in such settings, it is inequality of international best practice to use consumption In a society in which opportunities are equal, outcomes is as a more reliable measure of household welfare. a girl (let’s call her Goma) born to a poor and But consumption is also “smoother” across the objectionable illiterate Dalit couple in rural Kalikot should population and hence likely to understate in- have an equal shot at becoming a doctor, or an only to the equality. So the question of which measure of engineer as a boy (let’s call him Avidit) born in extent that it inequality should be used to characterize overall a well-educated, affluent and upper caste house- is caused by inequality becomes a tricky one, especially in a hold in urban Kathmandu. And so should a girl case like Nepal where the conclusions one might inequality of (let’s call her Champa) born to parents of other- come up with may be drastically different de- wise similar characteristics as Goma but living in opportunity. pending on that choice. rural Sarlahi. But how equal/unequal are basic 31 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L opportunities for children such as Goma, Avidit because family, society, geography or the govern- and Champa in Nepal? There may be disagree- ment should be responsible for ensuring whether ments on what to do about the social and eco- or not she will have access to them.11 While the nomic processes that led to their parents’ being outcomes in life also depends on genetic traits where they are in the current socio-economic that a child inherits from parents, most would ladder of Nepal. But everyone will agree that agree that the critical inputs necessary for the that position should not determine the chil- child’s development should be provided to all dren’s life chances. children regardless of his/her genetic make-up. Measuring inequality of Interventions to equalize opportunities earlier in opportunity life are also found to be significantly more cost- Measuring opportunities is harder than measur- effective and successful than are those attempted ing outcomes. This is partly because the defini- later in life. Research shows that preschoolers tion of opportunity is hard to concretize. What with low levels of cognitive development have constitutes “opportunities”? The answer may lower school achievement and earn lower wages vary from person to person. In order to simplify in adulthood.12 More recent studies suggest that matters, for this work, we define “opportunity” early childhood education has substantial long- as access to basic goods and services, in educa- term impacts, ranging from adult earnings to re- tion, health and basic infrastructure that are tirement savings.13 Moreover, lost opportunities necessary for an individual to realize his/her hu- during childhood cannot always be compensated man potential. for. Child malnutrition, for example, can gener- ate life-long learning difficulties, poor health and When opportunities are defined in this manner, lower productivity and earnings over a lifetime.14 the set of opportunities affecting outcomes later in life can be infinite, from access to primary With opportunities defined in this manner, the education and nutrition when very young to ac- inequality in the access to these opportunities Malnutrition cess to credit and employment opportunities as will be measured by the extent to which “cir- an adult. Although theorists frequently disagree cumstances” – or characteristics that are pre- during on where to draw the line, there is widespread determined at birth and characteristics that the childhood agreement that at the very least, basic opportu- child cannot and should not be held account- can generate nities should be affordable (otherwise universal able, e.g., gender, caste/ethnicity, place of birth, learning access would not be economically feasible) and parental position in the socio economic ladder completely outside the control of the individual. etc. – are correlated with the access to these op- difficulties, A reasonable interpretation of the first condi- portunities. In a true equal opportunity society, poor health tion is that universalization of a good or ser- these circumstances will exert zero influence in and lower vice is technologically feasible given the state of the access to opportunities. The higher the cor- productivity knowledge and economically viable, at least in relation between circumstances and access (i.e., the medium term. The second condition neces- if being born a boy makes a child more likely and earnings sarily limits the set of opportunities, because it to have an opportunity than being born a girl), over a excludes any good or service that is available to the more skewed and unequal the distribution lifetime. an individual as a result of effort or ability. of opportunities within the population. For the purpose of this work, we also remove Deep inequality of basic the ambiguity of this definition by limiting the opportunities space of opportunities to those that are provided to Using this basic conceptual structure and apply- an individual in childhood, a stage of life during ing a methodology that has now become fairly which the individual cannot be held responsible standard in the literature to a variety of datas- for his/her actions. Adults may make choices ets, we report deep inequality of opportunities that inhibit their own opportunities, such as in Nepal.(See Annex 1 for a simple exposition dropping out of school due to lack of effort. For of this methodology)Basic opportunities such as a child, however, opportunities are exogenous, starting primary school on time, finishing pri- 32 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 13. CIRCUMSTANCES ASSOCIATED WITH INEQUALITY OF BASIC OPPORTUNITIES Electricity Adequate sanitation Clean water Not stunted Finished primary on time Started primary on time 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Gender Household Head Characteristics Urban/Rural Region/Belt Ethnicity Parental Wealth Source: World Bank Staff calculations based on Demographic and Health Survey Data, 2011 Note: This analysis is presented only for opportunities for which the total inequality of opportunity is higher than 5 percent. See Annex 1 for details on methodology. mary school on time, escaping the scourge of tain kind of school but not learn the material ad- chronic malnutrition in early childhood, having equately enough to progress through the grades clean water to drink, clean and adequate sanita- in the timely manner and eventually acquire the tion facilities and electricity at home appear to skills rewarded in the labor market. Similarly a be far from universal and unevenly distributed child may live in a household connected to a tap even among those who have it. that is never serviced. There is quite a bit of variation in the circum- In a study of equality of opportunity, it is im- stances that contribute the most to the inequali- portant to assess what would be the minimum ties, but parental wealth, place of birth (whether threshold of quality for each basic good or ser- it is urban or rural, in the mountains, hills or te- vice. For example, sanitation may be considered rai, or if it is the central and eastern region or the a basic good, but a researcher must determine mid and far western region) and other parental the level of sanitation that can be deemed a characteristics (such as age, gender and educa- minimum standard within a particular cultural tion) appear to account for most of the observed context. In case of education, a true measure of inequality. (Figure 13) The strong association quality would be the quality of learning. But between parental characteristics and opportu- learning is difficult to measure and data on prox- nities for Nepali children suggests a significant ies such as test scores are hard to come by. Im- divergence in life trajectories very early in life. perfect and indirect variables such as completion The divergent opportunity profiles for Goma, of primary school on time (which may be related Champa and Avidit shown in Box 1 underscore to learning outcomes or student achievement as the point that being born to the right parents it requires timely grade progression) may have to and being born in the right place – both factors be used. For Nepal we use data on performance determined essentially by the lottery of birth – in the School Leaving Certificate (SLC) and remain the key determinants of life chances for National Assessment of Student Achievement children in Nepal. (Box 1) (NASA) exams administered on eight-graders across the country. Opportunity gaps wider still, if quality taken into account The grading system of the SLC examination has Basic goods and services are usually not homo- been reformed very recently. But this examina- geneous and there are large variations in quality. tion, long considered the “iron gate” of the Ne- Not all schools, for example, provide the same pali educational system, has been notorious for quality of instruction. A child may go to a cer- low pass rates. Using data collected in 2005, we 33 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L find that not only are pass rates low, they are national Student Assessment (PISA). The under- skewed in favor of children with favorable birth lying “opportunity” in our analysis using this data circumstances. Children from private schools, is whether or not the child has learnt the material with better educated parents, in schools located well enough to pass this eighth grade examination in district headquarters, and of the favorable in science, mathematics and Nepali. gender (male) are more likely to pass the exam as well as score marks deemed in the first divi- Based on the results in Figure 15, the highest sion. (Figure 14) inequality appears to be in the pass rate for mathematics. Among the factors that contribute Disparities also show up prominently in the odds to the inequality, school type is the most impor- of passing the National Assessment of Student tant, followed by geographical variables captur- Achievement (NASA) examination administered ing the region/belt/district of the birth, paren- to eighth graders across Nepal. NASA is the na- tal education and occupation. Relative to these tional equivalent of the global Program for Inter- circumstances, it is noteworthy that the caste/ BOX 1: CONTRASTING OPPORTUNITY PROFILES OF THREE CHILDREN IN NEPAL Goma is a girl, born in rural Kalikot. Her parents are entrepreneurs and businessmen. But in Nepal, the illiterate, belong to the Dailit community and are in life trajectory of these children begins to diverge the bottom 20 percent of Nepal’s wealth distribution. very early in life. Goma – whose life chances appear Champa is also a girl born to a household otherwise marginally better than Champa’s – is significantly very similar to Goma’s. But Champa’s parents are more likely to start primary school late, drop out of from a village in Siraha. Avidit is a boy born to an school, be stunted and grow up with limited access upper caste household in urban Kathmandu. Both his to basic amenities such as clean water, sanitation parents have a university education and come from and electricity than Avidit. Even in the 40 percent affluent backgrounds. chance that Goma and Champa complete primary school and go on into secondary and tertiary In a society where opportunities are equally available education, the quality of the education they receive for children of all socio-economic backgrounds, is likely to be vastly poorer than what Avidit will Goma, Avidit and Champa would all have equal receive in Kathmandu. odds becoming doctors, or engineers or successful 100 100 95 96 95 90 87 85 84 84 84 80 80 70 69 70 60 53 50 47 48 48 44 39 40 36 31 30 20 9 10 6 6 5 0 Started primary Finished primary Fully immunized Exclusively breastfed Not stunted Clean water Adequate Electricity on time on time sanitation Avidit Goma Champa Source: World Bank Staff estimates based on data from DHS (2011) 34 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 14. INEQUALITY IN SCHOOL LEAVING CERTIFICATE EXAMINATION (2005) (a) Coverage and equity adjusted coverage or HOI (%) 8 First Division 17 43 Pass 51 0% 10% 20% 30% 40% 50% 60% HOI Coverage (b) Contributions to Inequality (%) First Division 5 11 17 57 Pass 9 14 19 45 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Ethnicity Family Income Gender District HQ Parental Education Private/Public Source: World Bank Staff calculations using data from SLC Examinations Survey (2005), Ministry of Education, Government of Nepal Note: HOI, or the equity adjusted coverage, measures inequality of opportunity. The closer it is to 100, the more universal the opportunity is, i.e., the higher the fraction of population that “have” the opportunity. The closer this number is to the “coverage” the lower the inequality among those who have it. See Annex 1for details on methodology. ethnicity of students have a relatively small con- larly, what about inequalities in the marketplace tribution to the total inequality. Interestingly, for jobs? Focus group discussions conducted in the language spoken at home (whether or not the course of qualitative research done for this it is Nepali) accounts for a non-trivial amount work revealed a significant fraction of partici- of inequality in pass rates for Nepali language pants considered inequality in access to produc- (10 percent) and science (6 percent).Language tive employment as the most significant handi- does not seem to matter for mathematics. Given cap to their aspirations for economic mobility. that the medium of instruction in public schools Opportunities were perceived to be limited and is almost always Nepali, this points to the pos- skewed in favor of those with “strong politi- sibility of non-native Nepali speakers (such as cal connections” as opposed to being based on Madhesis, for example) having a disadvantage merit. in subjects that require free form expressions as opposed to those that require the solving of nu- In principle, the same methodological appara- merical and logical problems. tus could be used to analyze the extent to which one’s birth circumstances influence access to Large inequalities also in job jobs. But before embarking on such an exercise, opportunities it is useful to delineate what exactly we mean by What about opportunities beyond those that re- labor market opportunities. In the case of chil- late to the acquisition of human capital? Particu- dren, opportunities were defined as access to ser- 35 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 15. INEQUALITY IN PASSING EIGHTH GRADE STUDENT ASSESSMENT EXAMINATION (c) Coverage and equity adjusted coverage or HOI (%) Science Nepali Math (d) Contributions to Inequality (%) Source: World Bank Staff calculations using data from NASA Survey (2013), Ministry of Education, Government of Nepal Note: HOI, or the equity adjusted coverage, measures inequality of opportunity. The closer it is to 100, the more universal the opportunity is, i.e., the higher the fraction of population that “have” the opportunity. The closer this number is to the “coverage” the lower the inequality among those who have it. vices such decent quality of education, adequate nities” – in the purest sense – as well as what health, clean drinking water and sanitation etc. an individual makes out of those opportunities These were fundamental building blocks for the through own hard work and enterprise. Unlike development of their human capital and it can for children, the realm of individual agency and be claimed that most societies would agree that responsibility is potentially larger for individuals all children – irrespective of the circumstances on the verge of entry into the labor market. So determined by birth – should be provided with while there is certain appeal of the universality these basic opportunities. to opportunities such as education when they are applied to children – the statement that all But extending that idea to labor market is slight- children should have access to primary schools, ly complicated. Consider the most basic maker for example, is generally acceptable in most soci- of a labor market opportunity: being employed eties – the same cannot be said for employment in any kind of a job. Can this be regarded as an and other labor market related indicators. The opportunity? Any observed labor market state closest one can come is perhaps not that every- is a function of a set of accumulated “opportu- one should have a job, but that everyone who is 36 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 16. INEQUALITY IN LABOR MARKET OPPORTUNITIES, 2014 (a) Coverage and equity adjusted coverage or HOI (%) (b) Contributions to Inequality (%) Permanent job 29 42 11 13 Public sector job 23 48 11 15 Salaried job 3 55 30 0% 10%2 0% 30%4 0% 50%6 0% 70%8 0% 90% 100% Age Ethnicity District of Birth Urban/Rural Source: World Bank Staff calculations based on the sample used for the Perceptions of Poverty, Prosperity and Economic Mobility Survey implemented in 2014. Note: HOI, or the equity adjusted coverage, measures inequality of opportunity. The closer it is to 100, the more universal the opportunity is, i.e., the higher the fraction of population that “have” the opportunity. The closer this number is to the “coverage” the lower the inequality among those who have it. suitably skilled/ qualified or adequately experi- a universal “good” that is often encapsulated in enced should have a decent shot at getting a job an “opportunity” is violated in the case of jobs. of befitting his/her skills and expertise. Another way to see this is to recognize that “hav- Even with this qualification and accounting for ing a job” is essentially an equilibrium phenom- the role of other factors such as schooling, edu- enon that encapsulates individual motivations, cation, skills and experience in the state of being choices, and decisions some of which we can employed (in a job of any kind), it is actually observe in the data and control for and many of not always obvious that such a state is always a which we can do nothing about. If we wanted more desirable state. For example, an individual to seriously apply the inequality of opportunity who is unemployed at the moment because he analysis, perhaps the appropriate “opportunity” can afford to wait for a better job (e.g., queu- to use would be employability. The extent to ing for public sector jobs) is decidedly better off which employability depends on circumstances than an individual without a job because none beyond ones control would be the question of is available even though they would both appear interest. However, employability is hard to ob- identical in the data. Furthermore, the former serve and measure in typical labor force/house- may also be better off than somebody who ap- hold survey. pears to be employed but is in fact doing a job of drudgery and doing it simply because there is no To get around these issues, first, in addition to other alternative. So in some sense, the idea of the standard set of circumstances, we include in- 37 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 38 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L dividual characteristics like education and age to for close to a half of the total inequality in all Access to account for the employability of the individual. three opportunities. This is not entirely undesir- educational The exercise then is to determine the role of cir- able because education should be an important cumstances on labor market variables, net of age discriminant in a well-functioning, meritocratic opportunities is and experience. Note that by doing this we only and fair labor market that values skills. In fact, itself unequal capture the direct contribution of circumstances ideally we would want education (to the extent and there are to inequalities in the labor market variables and that it is a good measure of skills) to be the main important do not account for the role of circumstances of inequality. But as documented earlier, it is through their effect on earlier human capital ac- important to recall that access to educational advantages for cumulation, or the indirect channel. Second, in- opportunities is itself unequal and there are im- individuals with stead of using the basic state of being employed as portant advantages for individuals with favor- favorable birth the relevant metric, we use a finer, more restric- able birth circumstances. Which means that if circumstances. tive definition of labor market outcomes. This is we took into account the indirect effect of the to keep the focus on outcomes for which the state role of birth circumstances on labor market of realization of these outcomes is unambiguously outcomes – through their effect on the level of better than the state of not having them. In com- educational attainment – the size of the unfair parison to being employed, the state of being em- inequality in all three of these indicators would ployed in a job with certain characteristics such as be larger than what is currently shown. a job with a contract, or a job in the public sector is arguably more desirable. Among birth circumstances, father’s educa- tion and occupation appear to play the most So to adhere to the “opportunity” language, important role in the labor market inequalities those with an opportunity are those with jobs analyzed here. The results suggest that children of these desirable characteristics while those of better educated fathers and fathers in better without the opportunity are those without jobs forms of employment, have stronger chances with the desired characteristics, including those of having these good jobs. Caste/ethnicity ap- who are unemployed. The specific labor market pears important in the case of permanent or variables we use are as follows: (a) employed in a public sector jobs but not so much for the salaried job; (b) employed in a government job; general class of salaried jobs. Likewise, age of and (c) employed in a permanent job.15 the individual contributes between 23 and 29 percent of the total inequality in public sec- First, 27 percent of those in the labor force were in tor and permanent jobs respectively, while the some form of salaried employment and roughly a contribution to the inequality in salaried job third of all in salaried employment were either in is miniscule (3 percent). In contrast to other public sector jobs or permanent jobs. (Figure 16) opportunities related to human capital devel- As before, the HOIs for each of these indicators opment for which geographic attributes of the are lower than the coverage, suggesting some in- place of birth (urban vs rural, terai vs. hills vs. equality by underlying characteristics. mountains etc.) played an important role, it is interesting to note here that place of birth ap- In terms of key drivers of the inequality, edu- pears to account for very small shares of total cation of the individual stands out, accounting inequality in access to these jobs. 39 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 40 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L III. ECONOMIC AND SOCIAL MOBILITY Sensitivity to inequality is amplified in societ- ity in occupation than in education: 47 percent ies that lack mobility. Conversely, societies with of the overall population remained in the same high degree of mobility are able to heal the occupation as their father while about 43.6 per- wounds of structural inequalities. In an inter- cent of the population realized upward mobility generational sense, when children can aspire to in terms of occupation. (Table 3) achieve and indeed achieve levels of education, jobs, and living standards that are materially dif- Prospects of upward mobility ferent from the levels enjoyed by their parents, have improved over time inequality begins to decline over time. Mobility Prospects for upward mobility have been in- is important also from the perspective of dy- creasing in Nepal for both education and occu- namic efficiency. A mobile society is necessarily pation. (Figure 17) Point estimates suggest that also a meritocratic one and such a society is bet- while the likelihood of being better educated ter able to mobilize and utilize talents from all than one’s father was below 50 percent for those segments of the population (including the most born before 1965, the number is closer to 70 disadvantaged). This is good for growth. percent for the cohort born between 1985 and 1995, the youngest cohort in the sample. The Better educated than fathers, pattern is similar for occupation as well except but not necessarily better there is a discontinuous jump up (roughly 10 employed percentage points) in predicted probability of Analyzing intergenerational mobility in the space upward mobility for the 1975-1985 birth-co- of income requires data that is hard to get, not hort in comparison to the earlier cohorts. One just in Nepal but in most other developed coun- possible explanation for this is that in contrast tries as well. But we attempt to make some prog- to the earlier cohorts, this group of children ress on this here using educational attainment would have been in school around the time Ne- and occupation, two strong proximate correlates pal’s economy started liberalizing (1990-1994) of income. Defining upward mobility as attaining and thus may have been in a favorable position a higher level of education than one’s father or to seize the opportunities created by the result- being in a better occupation, we see that about 62 ing diversification of the economy. In any case, percent Nepalis (above the age of 30) experienced younger cohorts have been more upwardly mo- upward mobility in education. In comparison, bile than the cohorts born before 1975 who ap- about 29.8 percent stayed at the levels of their pear to have a low steady state transition prob- fathers and about 8.3 percent had levels of educa- ability of around 33-34 percent of doing better tion lower than their fathers. There is less mobil- than their father. TABLE 3: UPWARD AND DOWNWARD MOBILITY IN EDUCATION AND OCCUPATION, RELATIVE TO FATHERS Education Occupation (% of population > 30 yrs.) (% of population> 30 yrs.) Up 61.9 43.6 Same 29.8 46.8 Down 8.3 9.6 Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and Economic Mobility in Nepal, 2014 41 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 17: UPWARD MOBILITY BY AGE COHORTS But not for all social groups Consistent with the findings of high inequal- (a) Better educated than father ity of basic opportunities, prospects of upward mobility vary among the various social groups. .7 In particular, Newars, followed closely by hill Brahmins and Chhetris have the best odds of surpassing their father’s education and occupa- .65 tion levels. Nepalis in Muslim and other minor- Predicted probability ity caste categories followed by Terai Brahmin/ .6 Middle Caste, and Dalit categories have the lowest odds of upward mobility for education as well as occupation. Again, low sample size .55 for some groups, such as Muslims and Others, means that the standard errors around the point .5 estimates are fairly large so the results have to be interpreted with caution. For example, even though the point estimates for Dalits and the .45 Before 1965 Terai Brahmins and Middle Castes show slight differences, the 95 percent confidence interval (b) Better occupational status than father around these estimates have considerable over- laps suggesting that it is hard to draw any con- .55 clusions distinguishing the two groups. (Figure 18) .5 Rural-urban migration, ticket to upward mobility Predicted probability Prospects of upward mobility in education are .45 significantly higher for households that move to urban areas relative to who remain in rural areas of birth. (Figure 19) The difference in the prob- .4 ability of doing better than ones father between the two groups is close to 10 percent. This sug- .35 gests that physical mobility between rural and urban areas is a critical part of intergenerational mobility in education. The association between .3 rural-urban migration and upward mobility in Before 1965 occupation is even more pronounced: probabil- Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and ity of being in a better occupation than one’s Economic Mobility in Nepal, 2014 Note: The bars show 95% confidence interval father jumps by 28 percentage points (from 38 percent to 66 percent) for individuals who are born in rural areas but migrate to urban areas. This result is consistent with large gaps in op- portunities between urban and rural areas of Nepal. Poverty transitions show significant movement even within a lifetime Low correlation between parental socioeco- nomic status (income, jobs or education) and children’s outcomes is one hallmark of an inter- generationally mobile society. But how can mo- 42 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L bility within generations be measured and ana- FIGURE 18: UPWARD MOBILITY BY CASTE/ETHNICITY lyzed? One possibility is to consider directional income/living standards movement for the same (a) Better educated than father individual or household over a lifetime. A mo- .8 bile society would be one in which poorer and socially marginalized households can climb up .7 the economic ladder and attain higher living standards over a lifetime. In order to make “rank Predicted probability improvements”, the income of this group would .6 have to grow faster than the income growth real- ized by the average households. A key statistic to measure intra-generational mobility this way .5 would simply be the growth rate of income of the household between two points in time. .4 This kind of analysis however requires a nation- ally representative data that allows us to observe .3 the same household at two points in time. Un- Hill Brah. Hill Chhe. Terai B&MC Dalit Newar Hill Jana. Terai Jana. Musl.& Oth fortunately for Nepal, the panel data collected with the three rounds of the living standards (b) Better occupational status than father survey has suffered from a high degree of attri- .8 tion between rounds and is thus not ideal for this analysis. In order to overcome this short- coming, we use an innovative “synthetic panel methodology” that has been successfully applied .6 in similar settings in several other countries.16 Predicted probability (Annex 2 has details on this methodology) .4 Also, rather than analyzing directional move- ments throughout the income distribution, we focus primarily on two states: being in poverty and not being in poverty. This reduces dimen- .2 sionality and crystalizes the question of intra- generational mobility to the question of whether a person born poor can expect to climb out of 0 poverty in his lifetime or has to perennially re- Hill Brah. Hill Chhe. Terai B&MC Dalit Newar Hill Jana. Terai Jana. Musl.& Oth. main in poverty. Conversely, is there a possibil- ity that a non-poor individual might fall back Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and into poverty? Economic Mobility in Nepal, 2014 Notes: Caste/Ethnicity categories are hill Brahmin, hill Chhettri, Terai Brahmin and Middle Castes, Dalit, Newar, Hill Janajati, Terai Janajati, Muslim and Others. The current understanding of poverty is static and it is not always fully appreciated that the commonly reported poverty trends are net rates that obscure two crucial and contrasting phenomena of people falling into poverty and people escaping poverty. Those who fall into poverty and who are never able to escape pov- erty are distinct and the policy response neces- sary to address the needs of these two disparate groups is also different. For those who remain trapped in poverty, the chronic poor, enhanc- ing mobility may require focusing on expanding 43 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L access to economic opportunities. This includes direction and fell into poverty. (Figure 20) Like- opportunities to build assets (human capital) wise, between 2003-04 and 2010-11, 29 per- in childhood and opportunities to command a cent moved out of poverty while 14 percent fell return on these assets through a fair access to back in. This implies that for every two Nepalis the marketplace for jobs in adulthood. For those that got out of poverty in 2010-11, one slid back vulnerable to falling back into poverty on the into poverty during the second episode. There is other hand, what may be required is strengthen- also a sizeable persistence of poverty status over ing the coverage, reach and if needed also the the years. In both episodes, roughly three out generosity of safety net systems. of five Nepalis who were found to be poor in the latter period were actually already also poor Our analysis shows that between 1995-96 and in the earlier period. There are different ways 2003-04, 26 percent of the population escaped of thinking about conceptualizing the idea of poverty while 17 percent moved in the opposite chronic poverty, but if we consider households to be chronically poor (in the short run) if they FIGURE 19. UPWARD MOBILITY BY URBAN/RURAL MIGRATION STATUS are found to be poor in two consecutive survey rounds, separated roughly by seven years in the (a) Better educated than father case of Nepal, then we may conclude that about 60 percent of those that are found poor for any .75 given year actually suffer from chronic poverty. Those escaping poverty have largely clustered in a vulnerable state and despite a three- fold .7 increase in its proportional size, the size of the Predicted probability middle class remains modest .65 This concept of poverty transitions, in particu- lar the notion of the risk of falling into poverty, can be taken a little bit further to devise a more general definition of vulnerability. Like chronic .6 poverty, vulnerability is also widely acknowl- edged as a critical dimension of welfare in Ne- pal. But defining a broader class of vulnerabili- .55 Born rural lives rural Born rural lives urban Born urban lives urban ties or the size of the Nepali population facing these vulnerabilities has been difficult. Heuristic (b) Better occupational status than father constructs such as the size of the “population clustered just above the poverty” line have been .8 used to define and gain analytical traction on the vulnerable group. For policy formulation .7 and targeting purposes, social and demographic characteristics of the population (e.g., children, .6 Predicted probability elderly, widowed) have been the preferred mark- ers of vulnerability. .4 .5 As the recent earthquake has shown, there is a high degree of vulnerability to natural disasters .3 in Nepal. These vulnerabilities are heightened if one considers the increased frequency with .2 which these disasters are likely to occur due to climate change. Beyond catastrophic shocks .1 such as the earthquake that grab headlines and Born rural lives rural Born rural lives urban Born urban lives urban draw attention, a large proportion of Nepalis Source: World Bank Staff estimates based on data from Perceptions of Poverty, Prosperity and also face sharp variations in income due to more Economic Mobility in Nepal, 2014 44 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L mundane every day disasters such as variation in FIGURE 20. EPISODES OF MOVEMENTS IN AND OUT OF POVERTY monsoon rainfalls, floods, droughts, landslides, (% OF POPULATION) forest fires, illness and accidents. If credit mar- (c) Episode between 1995/96-2003/04 kets were to work well and if households were 35 33 able to insure against these shocks, one would expect consumption and living standards to not 30 be affected by much. But recent research sug- 25 26 24 gests that households in rural Nepal in particu- lar are from being shielded from such risks. In 20 fact, by some estimates, the level of exposure is 17 15 such that a monsoon rainfall that is 10 percent below the historical norm can lead to a 0.12 10 standard deviation decline in the height-for-age for children.17 5 0 Using the three rounds of the living standards Poor, poor Poor, nonpoor Nonpoor, poor Nonpoor, nonpoor data and applying the vulnerability of falling into poverty method, we define a “vulnerability (d) Episode between 2003/04-2010/11 line” as the level of consumption below which 40 a particular household has larger than 10 per- 36 35 cent probability of falling back into poverty.18 Households that are above the poverty line but 30 29 below this vulnerability line are categorized as 25 vulnerable. Why do we do this instead of an- choring the definition of vulnerability along a 20 20 more absolute concept, such as say, a multiple of 15 14 the poverty line as it is sometimes done? The rea- son is Amartya Sen’s(1983) argument that “pov- 10 erty is absolute in the realm of capabilities but 5 relative in the realm of functionings”. Extend- 0 ing that notion to vulnerability, this approach takes an absolute stance on what it means to be Poor, poor Poor, nonpoor Nonpoor, poor Nonpoor, nonpoor vulnerable (when a household’s probability of Source: World Bank staff calculations using data from three rounds of NLSS data for 1995/96, falling into poverty is larger than 10%) but a 2003/04 and 2010/11. The methodology used is called the synthetic panel methodology. “relative” position on how much income or con- sumption is required to achieve that capability. cluding India and Pakistan, Duflo and Banerjee (2009) propose from $2 to $4 per person per An additional benefit of this methodology is day as lower and upper thresholds respectively that it also allows us to get at the notion of the for the middle class. A global study of the mid- middle class. The vulnerability line, by defini- dle class conducted by the Asian Development tion, forms the “lower threshold” for the middle Bank defined the middle class in Asia (and also class population which gets defined as house- in Nepal) as those with consumption between holds with sufficiently low probability of falling $2 and $20 per person per day.19 In our case, into poverty. But what should the “upper thresh- the lower threshold is determined by the vulner- old” of this middle class should be? That is, what ability line which translates to roughly $3 per is the level of consumption that distinguishes person per day in 2014 in 2011 PPP terms. For the middle class from the top strata, the upper the upper threshold, we inspect the consump- class, or the elite? tion distribution and pick a value that comes to about $8 per person per day in 2014. Even Here the literature offers a number of sugges- though this choice somewhat arbitrary, it is be- tions. Based on their analysis of household con- tween the global upper thresholds of $4 and $20 sumption data from 13 developing countries in- per person per day. Besides, given how thin the 45 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 21. THE POOR, VULNERABLE AND THE MIDDLE CLASS IN NEPAL 96 to 45 percent in 2010-11. (Figure 21) During the same period, there has also been appreciable 100% 7 growth in the size of the middle class population 90% 14 22 in Nepal with the proportion increasing from 7 80% 28 percent in 1995-96 to 22 percent in 2010-11. 70% This implies that roughly 1 out of every 5 Nepali 35 60% today belongs to the middle class. 45 50% 40% The growth of the middle class is widely regard- 30% 64 ed as the bedrock for the development of stable 20% 50 and prosperous societies. The middle class group 31 is typically associated with high entrepreneurial 10% activity, high investment in human capital, and 0% 1995 /96 2003 /04 2010 /11 progressive political and economic value sys- tems.20 The middle class group often also acts as Poor Vulnerable Middle Class Upper a neutralizing force between the extractive ten- Source: World Bank staff calculations using data from three rounds of NLSS data for 1995/96, dencies of the elites and the revolutionary ten- 2003/04 and 2010/11. The methodology used is called the synthetic panel methodology. dencies of the poor. As such, societies with large and diverse middle class are often stable societ- consumption distribution is at that level, the ef- ies. The middle class is a source of demand for fect of the choice has minimal impact on the size consumer goods and services and can provide of the middle class. In other words, the size of much needed succor for the development of do- the middle class is much more sensitive to the mestic industries not just in manufacturing but choice of the lower threshold than the upper also in high value services such as tourism and threshold. banking. It can also be a robust voice in the de- mand for good governance and the provision of Using this method, we see that as the proportion quality services which will indirectly also benefit of the poor has declined over the years, there has the poorest. Unlike the strugglers or those in the been a corresponding growth in the size of the vulnerable category, a secure middle class can af- vulnerable population from 28 percent in 1995- ford a longer planning horizon and is thus more FIGURE 22. GEOGRAPHIC CHARACTERISTICS OF THE POOR, VULNERABLE AND THE MIDDLE CLASS, 2011 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Rural 35 48 17 Urban 13 35 47 0 Eastern 30 51 18 Central 22 44 31 Western 29 46 24 Mid Western 45 45 10 Far Western 55 33 11 Poor Vulnerable Middle class Upper class Source: World Bank staff estimates based on data from three rounds of NLSS data for 1995/96, 2003/04 and 2010/11 46 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 23. THE POOR, VULNERABLE AND THE MIDDLE CLASS IN NEPAL Source: World Bank staff calculations using data from three rounds of NLSS data for 1995/96, 2003/04 and 2010/11. The methodology used is called the synthetic panel methodology. willing to trade off private costs in the short run significantly more likely to be poor or vulner- for reforms that will yield better public and col- able. Middle class households appear to be gen- lective goods such as quality schools, healthcare, erally in non-agricultural forms of employment roads and the environment. and the likelihood of being in the middle class is highest among Nepalis who are in salaried So who are the middle class in Nepal? Close to professional employment or in other forms of half of the urban population is in the middle self-employed services. Vulnerability is highest class while half of the rural areas is in Nepal is among those who are self-employed in agricul- in the vulnerable category. Middle class preva- ture and in other service sector employment, lence is the highest in the Central Development most likely in blue collar occupations. Over- region while the Mid and Far Western Regions all, poverty and vulnerability is highest among have the highest poverty and vulnerability. Vul- households deriving their livelihoods from wage nerability in the terai is higher than in the hills employment in agriculture. (Figure 23) and the mountains and since the terai is also more populous, the number of vulnerable is also Insecure and struggling higher there. Chances of being in the middle middle class class are highest in the hills and lowest in the Gallup World Survey asks respondents ques- mountains but the absolute size of the middle tions about their perceived position in an 11- class is again, larger in the terai owing to its step ladder denoting various parts of the income higher population. (Figure 22) distribution within each country. Respondents are asked about current position as well as posi- Looking at education and employment charac- tions five year before the survey and expected teristics, the likelihood of being in the middle position five year hence. Based on responses class is significantly higher for households head- to these three questions, Gallup defines house- ed by those with more than 11 years of educa- hold as thriving, struggling or suffering. Thriv- tion. Households with uneducated heads are ing households have strong current life situa- 47 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 48 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L tion and have positive views about the next 5 FIGURE 24. THE PROPORTION OF NEPALIS STRUGGLING, years. Struggling households have wellbeing SUFFERING AND THRIVING that is moderate but inconsistent while suffering 0.8 households have well-being that is at high risk, 0.7 0.7 they report poor ratings of their current life and also poor outlook for the future. Based on cross 0.6 country analysis of these categories, Gallup has 0.5 found thriving respondents to have fewer health problems, less worry and stress, sadness and 0.4 anger and more happiness, enjoyment, interest 0.3 and respect. Struggling households on the other 0.2 hand report higher daily stress, worry about 0.2 money than thriving respondent. In contrast, 0.1 0.1 suffering households are likely to be lacking in basic food and shelter and report to be under 0.0 physical pain, stress, worry and sadness. They Struggling Thriving carry more than double the disease burden as “thriving” respondents within each country. Source: World Bank staff estimates based on data from Gallup World Survey. Using data from Gallup for Nepal, we find that ing in the middle class with being engaged in 90 percent of Nepalis fall in the suffering or relatively secure professions such as government struggling categories. In contrast, only 10 per- jobs, teaching, medical and engineering profes- cent of the population actually falls in the thriv- sions, and other forms of private sector salaried ing category. (Figure 24) Juxtaposing this with employment. This association was reinforced the earlier result on the size of the middle class, by widespread recognition among respondents it appears that even though there is a nascent that economic prospects of those in this group middle class building up in Nepal, roughly half hinges more critically on professional skills and of this group is fundamentally insecure about labor as opposed on things like inherited capital the prospects of realizing continued improve- and other forms of financial assets. This view is ments in its living standards. consistent with what has been noted in the lit- erature on middle class.21 On the whole, these In the qualitative survey carried out for this results suggest that even though a majority of analysis, respondents were asked about what Nepalis today may not be in poverty directly, they saw as key characteristics of the middle they are either at risk of falling back into pov- class in Nepal. Both in depth interviews and erty or struggling to attain the level of economic focus group discussions, an overwhelming ma- security required to be comfortably ensconced jority of respondents/participants associated be- in the middle class. 49 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 50 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L IV. MIGRATION AND OTHER PATHWAYS OF MOBILITY Nepal has made tremendous strides in moving By 2011, overseas migration rates increased al- A majority of people out of the poverty line. But the same suc- most everywhere. Except for about 9 percent of Nepalis today cess has not been achieved in improving in liv- the villages, all villages experienced an increase ing standards sufficiently enough to move those in migration rates. The average increase was constitutes a who have escaped poverty to a more secure mid- about 4 percentage points per village and there vulnerable dle class. As a result, a majority of Nepalis to- is considerable variation in the increase. Villages group that is day constitutes a vulnerable group that is peril- with no migrants abroad accounted for less than perilously close ously close to falling back into poverty. Building one percent of all villages; share of villages with prosperity for a majority of Nepalis will entail migration rates higher than 10 percent rose to to falling back not just safeguarding and cementing the gains 30 percent. Migration increased the most in into poverty. already made, but also boosting productivity of Eastern and Western regions but again, there is the economy in a manner that will help them re- considerable heterogeneity within the regions. alize higher income levels. Work related migra- The increase in the spatial extent of internation- tion, primarily to overseas destinations, has been al migration has also deepened the penetration a critical driver of mobility experienced by Ne- of remittance flows which accrue directly to the palis over their lifetimes and across generations. households with migrants. This has resulted in a But one key limitation of the welfare improve- spatial transition characterized by greater rural- ments that have been realized of this process has urban migration, expansion of conventional cit- been that these private gains have not translated ies and smaller towns and fast growing urban into an aggregate improvement in the produc- sprawls around main highways bordering India tive potential of the economy in a manner that and most district headquarters. would enable the creation of good jobs in a large enough scale for those that remain in Nepal. For households not directly able to participate in international migration due to cost constraints, Spatial extent of migration or to even go to India, rural-urban migration and consumer cities still represents an important strategy for income One fact about the migration experience in Ne- diversification. Urban areas, particularly urban pal that is perhaps a little under-appreciated is Kathmandu, appear to have clear economic ad- its overall spatial extent. Be it the statistic about vantages from the point of view of living stan- the daily flow of migrants over a given period of dards. Decompositions of spatial disparities in time or the overall existing stock, migrants in- welfare suggest sizeable returns to these loca- creasingly come from everywhere in Nepal. For tions. (See Table 4) This is unsurprising given example, in 2001, an average village had the mi- the fact that urban areas have historically had gration to population ratio of about 4 percent, better services, infrastructure and connectivity with one-seventh of villages (14 percent) having relative to other parts of the country. But the ur- zero migrants abroad. Only 15 percent of the ban areas also provide better education, health- villages had migration rates higher than 10 per- care and other services and amenities the increas- cent of the resident population. The top panel of ingly better off rural households have reason to Figure 25 shows the distribution of village level value. Qualitative research undertaken for this migration rates in 2001. Though the districts in work suggests that households benefiting from the Western and Far-Western hills had high mi- migration opportunities abroad are not only gration rates, there was considerable heterogene- remitting cash but also preferences and tastes ity within different regions in the country. for certain lifestyle that is more likely to be ser- 51 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 25. INCREASING SPATIAL EXTENT OF MIGRANT POPULATION BETWEEN 2001 AND 2011 (A) 2001 (B) 2011 0.0 0.00 - 0.01 0.01 - 0.05 0.05 - 0.10 0.10 - 0.15 0.15+ Source: Shrestha (2015), background paper commissioned for this work. The source data is 2001 and 2011 Population Census. 52 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L TABLE 4: SOURCES OF WELFARE DISPARITIES BETWEEN VARIOUS GEOGRAPHIC AREAS OF NEPAL, 2011 Difference in Endowment Returns   log welfare (characteristics of the (characteristics of the ratio household, %) place %) Urban vs Rural 0.66 61 31 Kathmandu vs Mountains and rural Mid and Far Western Hills 1.20 47 53 Kathmandu vs Rural Terai 1.02 47 53 Kathmandu vs Other Urban Areas 0.58 45 55 Source: World Bank staff calculations based on data from NLSS III. Note: The endowment variables included in this decomposition exercise are household size, household demographic composition, dependency ratio, household head age, gender and education, spouse of the household head age, and education, and household head ethnicity. Welfare ratios expressed in logs. viced in urban areas. The growing amenity value Real wages have grown fairly evenly across Ne- placed on conveniences more readily available in pal between 1995 and 1996. But taking into urban areas, including essential basic services, as account cost of living differences across the na- a result of income growth in rural areas has also tion, the rate of wage growth in Kathmandu ap- been an important driver of city growth in Ne- pears to be more modest in comparison to wage pal. Indeed, one recent study on the drivers of growth witnessed in other parts of the country. domestic migration in Nepal corroborates this. This hints at the possibility of an erosion of The specific finding is that domestic migrants in the competitive allure of Kathmandu. (Figure Nepal prefer areas that are closer to their places 26) From the point of view of the producers, it of birth and have higher income but they also could be argued that the cost of living differenc- prefer areas with better access to public goods.22 es should not matter for decisions on where to Considered together, they conclude that access locate. Firms producing goods that are tradable to better amenities trumps the role of income in nationally, should, in principle, be undeterred decisions to migrate domestically. by high nominal wages especially if productiv- ity is high enough. In reality, however, the ex- But analysis of wages suggests that productivity tent to which the urban core of Kathmandu, if growth in urban areas has not kept pace with the not some of the other urban areas, is producing rising cost of living and thus threatens to un- tradable commodities is questionable. In all like- dermine the potential benefits of agglomeration. lihood, the urban economy is heavily oriented The theoretical literature on new economic ge- towards the non-tradable service sector which ography posits that concentration of economic does not bode well for the agglomeration poten- activities in cities has the potential to bring to- tial for Nepal’s largest city. gether ideas, capital, workers and consumers in a way that allows firms to defray costs of produc- Where the positive effects of agglomeration have tion across a larger body of consumers and real- remained elusive, the negative effects have be- ize productivity advantages. Cities also provide come prominent and visible. Unplanned urban better developed infrastructure, a high degree development and patchy enforcement of regula- of market specialization, greater competition, tions have led to a haphazard and uncontrolled information exchange and more efficient labor urban sprawl in and around Kathmandu and market matching. This allows firms located in other major cities. In several parts of the city, these urban clusters to benefit from the higher buildings stand so close to each other that it is density of the population by realizing increas- impossible even for emergency vehicles (ambu- ing returns in their production processes, or lances, fire brigades) to pass. Some of the newer “agglomeration economies”. When this process apartment buildings are built taller than the takes place, workers are able to enjoy growing maximum heights that the city’s fire brigades wages despite increasing inflow of new workers can reach. Though Kathmandu’s old city and to the cities seeking economic opportunities.23 some of the newer construction suffered heavy 53 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 26. SPATIAL VARIATION IN CHANGES IN REAL WAGES IN ance of their daily water requirements them- VARIOUS REGIONS OF NEPAL, 1995-2010 selves, through the extraction of groundwater. 10 As the primary seat of the government, the urban primacy of Kathmandu has historically 8 been guaranteed. But despite being the fastest growing metropolis not just in Nepal, but in all % per year of South Asia, Kathmandu’s share of Nepal’s to- 6 tal urban population has declined over the last two decades. In 1991, Kathmandu accounted 4 for 25 percent of Nepal’s urban population. By 2011, this number was down to 22 percent. 2 Considering Kathmandu valley as a whole, i.e., municipalities of Kathmandu, Lalitpur, Bhakta- 0 pur, Kirtipur and Madhyapur Thimi taken as a Kathmandu Other Urban R-W Hills R-E Hills R-W Terai R-E Terai Kathmandu Other Urban R-W Hills R-E Hills R-W Terai R-E Terai single urban cluster, the decline is even sharper and goes from 39 percent in 1991 to 32 percent in 2011. National price index, 2010 =1 Spatial price index, 2010 =1 Source: World Bank staff estimates based on data from Gallup World Survey. The rate of the decline could be slower if the recent conversions of large VDCs into munici- damages after the earthquakes in 2015, it has to palities is taken into account. In 2014 and 2015, be said that the level of destruction could have the government conferred municipality status to been significantly worse.24 several VDCs or cluster of VDCs, taking the number of municipalities from 58 during the The urban environment has also degraded sub- 2011 Census to 214 and several of these new stantially, eroding livability. Yale’s Environmen- municipalities are inside the Kathmandu val- tal Performance Index (EPI) for 2014 ranked ley and would technically become part of the Nepal 177 out of 178 countries on air quality. greater Kathmandu metropolis. But the general During some months (especially in the winter), point that there has been considerable growth of Kathmandu’s air quality deteriorates to alarming secondary towns and cities outside of the largest levels as brick kilns, diesel vehicles, generators urban core remains valid. and pump sets are fired up not just in the city proper, but all over the Indo-Gangetic plains Atypical structural pushing the emissions north towards the Hima- transformation layas. Another pressing environmental challenge One of the primary reasons, the spatial transfor- is solid waste management, which, unlike air mation has not been able to generate the growth pollution has entirely local roots. Collection is that is typically associated with cities is that it low, and only two municipalities in Nepal dis- has been accompanied by a rather atypical path pose of waste in a sanitary landfill. In most cas- of structural transformation. Agriculture, which es, the main waste disposal sites are riverbanks, used to command a dominant share of the total depressed land and dumps, open pits, or tem- value added in the economy has declined to 34 porary open piles. The discharge of untreated percent. The share of industries rose and reached wastewater into water bodies and unmanaged a peak of 23 percent around the beginning of solid waste has turned prominent river system the conflict in 1996, but declined to a low 16 into open sewers and garbage dumps. Kathman- percent since then. In the context of shrinking du in particular also suffers acute shortage of agriculture and timid industries, services have drinking water and access to piped water has ac- grown in prominence and account for more tually decreased in the last decade. (Muzzini and than half of the GDP today. The typical story of Aparaicio, 2013). The population has responded structural transformation is one through which to the water shortfall by pumping out the bal- workers leaving low productivity agriculture in 54 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L favor of higher productivity jobs in manufac- appears to have decreased over the 15 year pe- turing agglomerate in cities which act as pro- riod with the decline in casual wage work mar- duction centers and engines for growth. In the ginally edging out the increase in regular wage case of Nepal, the economy has diversified away jobs. In the urban areas as well, the employment from agriculture over the long run, but it ap- composition has not changed that drastically. pears to have skipped the intermediate manufac- There has been an increase in self-employment turing phase and gone directly to services. This in agriculture with a commensurate decrease “premature graduation” from industries towards in the proportion of those engaged in off-farm services is a peculiar aspect of Nepal’s structural self-employment. Overall wage employment in- transformation and potentially the root cause of creased by 0.3 percent over the 15 year period the apparent disconnect between private income between 1995 and 2010. Casual employment growth and the formation of a robust middle decreased while regular wage employment in- class.25 creased by about 2.5 percent. Despite the dramatic shift in the production Excluding self-employment and focusing just on base of the economy, the change in employment wage jobs, we see in Figure 28 that the sectoral patterns have been relatively modest. A sizeable contribution of these jobs has changed quite a majority of rural Nepalis continue to be self-em- bit. The results show some interesting patterns. ployed in agriculture as their primary occupa- In urban areas, wage employment in agriculture tion. (Figure 27) Off-farm self-employment has has all but disappeared, declining from 21.6 roughly doubled from 4.5 percent in 1995 to percent in 1995 to 4.9 percent in 2010. Wage 8.1 percent in 2010. Overall wage employment employment in services has also seen a decrease FIGURE 27. PRIMARY SOURCE OF EMPLOYMENT OF THE ECONOMICALLY ACTIVE POPULATION IN URBAN AND RURAL AREAS IN 1995 AND 2010/11 (a) Urban, 1995 4.5% 4.6% (b) Rural, 1995 10.5% 80.4% 8.1% 5.4% 8.0% 30.9% 27.5% (c) Urban, 2010 (d) Rural, 2010 9.0% Regular waged workers 32.6% 78.6% Casual waged workers Self-employment - agriculture Self-employment - non-agriculture Source: World Bank staff calculations based on data from NLSS I and NLSS III. 55 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 28. SECTORAL COMPOSITION OF WAGE EMPLOYMENT IN URBAN AND RURAL AREAS, 1995 AND 2010 (a) Urban, 1995 16.1% (b) Rural, 1995 21.6% 1.5% 8.6% 53.8% 13.4% 10.5% 63.4% 6.4% 4.8% 4.9% (c) Urban, 2010 (d) Rural, 2010 27.2% 25.5% 33.9% 45.3% 3.5% 9.7% Agriculture 20.4% Mining, Manufacturing 15.0% 14.6% Construction Professional Services Source: World Bank staff calculations based on data from NLSS I and NLSS III though it still accounts for close to half of all in wages in real terms over this period in these wage employment. In proportional terms, pro- sectors, the lack of stability and informality im- fessional employment (salaried white collar plies that they do not have the security of regu- jobs) has increased three fold and employment lar wage jobs that are the fundamental building in manufacturing has also doubled over the 15 blocks of a strong middle class. In addition, year period. Share of wage jobs in urban con- looking into the skill content of jobs in regular struction has increased from 6.4 percent in 1995 and casual categories in both urban and rural to 9.7 percent in 2010. In rural areas, wage em- areas of Nepal, we see that (a) casual jobs are ployment in agriculture has declined by about predominantly of the unskilled nature and (b) 30 percentage points during this period and regular jobs have a larger composition of skilled most of this loss has gone in the increase in em- workers, defined in our analysis here as workers ployment in the construction sector – employ- with higher than secondary education. But it is ment share for which grew from 8.6 percent to also noteworthy that the skill content of both 20.4 percent – and services. Since wage workers types of employment has been stagnant over the in agriculture have typically been the poorest de- longer term. (Figure 28) mographic group in Nepal, this pattern shows that a bulk of Nepal’s recent poverty reduction Too much of a good thing? is likely to have come from the income growth The fundamental missing link between Nepal’s that this group has been able to realize. success on improving living standards, particu- larly at the bottom part of the distribution and However, an overwhelming majority of these reducing poverty on the one hand and its con- new jobs in construction are of the casual kind. current failure to move people more robustly While workers have benefited from the growth out of the red zone of vulnerability into a secure 56 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L middle class is the country’s failure to generate FIGURE 29. SKILL AND GENDER COMPOSITION OF employment and jobs in the productive sectors WAGE JOBS, 1995 AND 2010 of the economy. This has been difficult mainly 1995 because Nepal finds itself today on a growth 100 plateau, mired by low level of investment on ac- count of protracted political uncertainty on the 80 one hand and shifting drivers of growth on the other (World Bank, 2014). 60 Percent Nepal’s nascent industrial sector in the early 40 1990s was driven substantially by the export led growth in its garments and carpets industries. The deteriorating security environment after the 20 onset of the violent Maoist conflict in 1996 put a damper on this growth. The eventual rescind- 0 Regular Casual Regular Casual ing of preferential access to lucrative US markets Urban Rural under the Multi Fibre Arrangement in 2005 all but killed the garment industry as Nepal couldn’t 2010 maintain competitiveness in this segment in the 100 world market. This took a heavy toll on Nepal’s industrial sector which has not really recovered 80 since then. (Figure 30) The growth in services is encouraging but there are questions about 60 the extent to which it can catapult the country Percent into a higher growth trajectory. Unlike the ser- 40 vice sector in India for example, Nepal’s service sector is largely informal, fragmented and with 20 limited promise for productivity growth or spill- overs effects. The growth in the sector has, to a large extent, been fueled by remittances which 0 Regular Casual Regular Casual have grown to levels so high that it is unlikely Urban Rural to expect them to grow further. (World Bank, Skilled, male Unskilled, male 2014& 2016b). Skilled, female Unskilled, female Given these circumstances, overseas work and Source: World Bank staff calculations based on data from NLSS I and NLSS III remittances have provided a much needed safety valve for the economy. At the micro level, the flow of remittances has directly benefited mi- grant households. Even households without mi- construction. The latter in particular, has gener- grants have benefited from the growth in labor ated tremendous growth in rents to the owners income. On the one hand, wages in agriculture of capital within cities perpetuating a multiplier have gone up, possibly as a result of the tight- effect on the demand for goods and services. ening of labor supply. On the other hand, in- This has translated to the macro level where the crease in the demand for non-agricultural goods remittance receipts have been critical to main- and services has led to an increase in demand taining Nepal’s current account position, despite for non-farm labor and pushed up wages. In a widening trade deficit as the country imports the urban and newly urbanizing areas, remit- a growing amount of consumer goods ranging tances have been the critical source of liquidity from the most basic such as rice and petroleum fueling growth in banking and financial services, products to telecommunications equipment, restaurants and trade as well as real estate and luxury cars and parts. The overall surplus po- 57 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 30. GROSS VALUE ADDED, % OF GDP the face of declining crude oil prices. (Figure 31) While the strategy has served as a useful buffer 80 Conflict during turbulent times for Nepal, it cannot be the durable source of growth required to propel Nepal into the trajectory required to become a 60 middle income country with a vibrant middle class by 2030. 40 % of GDP In order to accelerate the creation of productive jobs, Nepal must put in place policies to rekindle 20 growth, particularly in its industrial/manufactur- ing sectors. This is because it is the only sector in which there is sufficient scope for scalable 0 1965 1975 1985 Year 1995 2005 2015 improvements in productivity and the ability to generate formal sector employment for the grow- Agriculture Industry Services ing number of Nepalis joining the labor force Source: WDI, 2016 every year. Being a small market, finding a com- petitive advantage in the global economy will be crucial. But stabilizing the political environment, sition in the balance of payments, largely due carrying out the necessary reforms to improve the to remittance inflows have been critical buffer investment climate and alleviating the crippling against global macroeconomic shocks. infrastructure deficit would be equally essential entry points. According to a recent Gallup survey, But it could be also argued that the safety valve only 25 percent of Nepalis expressed satisfaction has in a perverse way weighed down on Nepal’s with the availability of good jobs in the economy. development potential. First, the buoyant influ- (Figure 32) This is lower than most other regional ence remittances have had on the purchasing neighbors and significantly lower than countries power of a majority of Nepalis has inadvertently in the South East Asia region, such as Vietnam also resulted in the appreciation of the country’s and Cambodia. real exchange rate which has in turn eroded the competitive edge of manufacturing industries, ...but it will remain important further undermining the employment gen- Concurrently, it must also be recognized that eration potential of this sector. (World Bank, there is no level of growth Nepal can realistically 2013) Second, the steady outflow of young and generate in the medium term that will generate potentially restless youth and the cushion of re- so many jobs that Nepal is able to completely mittances could have also lulled policymakers overturn the flow of migrant workers. A more into not seeing the immediate urgency of key realistic outlook is that a steady state level of reforms necessary to unlock Nepal’s longer term overseas migration will remain a critical dimen- development potential. A stable, positive bal- sion of Nepal’s jobs environment in the medium ance of payment position and a strong fiscal po- to long run. For this reason, building a better sition maintained in part by the revenues from understanding of not just the migration process taxes on an ever growing base of imports have in general but also of what can be done to make been the bulwarks of Nepal’s overall macroeco- it work better so that successive cohorts of Ne- nomic position over the last decade. pali migrants have better information of oppor- tunities abroad, can become more productive Migration may slow down... and can have access to better quality jobs abroad But remittance driven growth is not likely to appears an important part of the jobs agenda. continue forever, especially given the already high base. Further, there are signs that the rate Several recent studies document widespread of migrant outflow is beginning to taper off as misinformation among migrants on various the major host countries show signs of stress in aspects of their eventual jobs, the associated 58 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 31. MONTHLY OUTFLOW OF MIGRANT WORKERS FROM NEPAL Earthquake 60000 50000 Monthly outflow of migrant workers 40000 30000 20000 10000 January/08 January/10 January/12 January/14 January/16 Source: Ministry of Foreign Employment Database Note: The solid line represents the trend obtained by applying the Hodrick-Prescott filter to the raw data. benefits and also the real and hidden costs of $1200 on average for a job abroad. This amount migration (Shrestha, 2016a; Shrestha 2016b). is more than 3 times of the national per capita There are also significant costs to matching mi- GDP. Commissions and fees for several layers grant workers to jobs and the credit markets that of brokerage services that link the worker to the could make the migration a productive invest- job constitute the largest share of the up-front ment does not work as smoothly as they could. cost borne by the workers. Returnees who have For example, large gaps exist between expected completed their previous contract expect to pay and actual wage schedules abroad. Median in- $1100, an amount not very different from first experienced potential migrant expects to earn time migrants. However returnees who still have $390 in a month; this is 30 percent larger than a work contract abroad expect to pay less than what he actually ends up earning. Potential mi- a third of that amount. This suggests that the grants are gravely misinformed about mortality cost of linking the worker to a job abroad dwarfs risks abroad and react adversely when a migrant other migration related costs (travel tickets, pa- from the same district dies abroad. Following a perwork, documentation, permits etc.). The death incidence, migrant flow from the same financing needs often lead households to take district to the destination drops but a small frac- on debt. Compared to households without mi- tion choose to migrate to a different destination. grants, households with migrants in some of the (Shrestha 2016a) more lucrative destinations are 12 percent more likely to have an outstanding loan and twice as Likewise, the up-front costs of migration are likely to have loans exceeding $500. On aver- not trivial. Potential migrants expect to pay over age, the loan sizes for households with migrants 59 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 60 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L FIGURE 32. SATISFACTION WITH THE AVAILABILITY OF GOOD JOBS IN THE ECONOMY 0.8 0.71 0.7 0.6 0.56 0.5 0.46 % 0.4 0.38 0.35 0.33 0.3 0.28 0.25 0.2 0.1 0.0 Bangladesh Cambodia pakistan Vietnam SriLanka Nepal China india Source: World Bank staff calculations based on data from Gallup. are 73 percent larger than those for households involving young migrants in Nepal. Correcting without migrants to these countries. Finally, some of the inefficiencies would enhance the de- these loans often come with average interest rate velopment impact of labor market opportunities of about 25 percent per year and the larger loans that many young Nepalis seek abroad. Together usually accompany collateral requirements. with efforts to generate jobs within Nepal, im- proving the labor market intermediation process These stylized facts suggest significant inefficien- for potential migrants could be an important cies in the labor market intermediation process policy agenda for the medium run. 61 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 62 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L V. DIRECTIONS FOR POLICY Nepal has made good progress on reducing pov- true). So working towards leveling the playing Building a erty over the last two decades. This report ana- field for all could move the debate forward be- strong middle lyzes the drivers of this progress and identifies cause what is a minimum criterion for fairness three key challenges to consolidating these gains and justice for the left happens also to be the class is vital for to translate strong progress on poverty reduction foundation for meritocracy for the right. Nepal’s efforts into an equally strong impetus for the formation to reach the of a robust middle class. A strong middle class is IMPROVING SERVICE DELIVERY FOR ALL, middle income critical for Nepal’s efforts to reach the middle in- AND PARTICULARLY FOR THE POOR IN come country status by 2030 and the analysis in LAGGING REGIONS country status this report points to three particular policy areas Operationalizing the idea of equality of oppor- by 2030 that could help the country get there: equalizing tunity into measurement, we document deep opportunity, reducing vulnerability and boost- inequality of opportunities in Nepal. These are ing productivity. manifest in the unevenness in the access to basic opportunities such health and education as well Equalizing Opportunity as jobs. Intergenerational mobility also appears Widespread perception of structural inequali- constrained for particular social groups and geo- ties and the increasingly vocal demand for so- graphic areas. Among other factors, education cial and economic inclusion have had a deep emerges be the strongest determinant of the ac- influence on Nepal’s recent history. The 10-year cess to the few good jobs in Nepal today. But long Maoist conflict, abolition of a 250 year old access to quality education remains the preserve monarchy, and a radical reconfiguration of the of those lucky enough to be born to the right state structure, are all direct or indirect conse- parents: parents who are themselves educated, quences of this. Even today, the issue remains living in cities and in certain regions, able to touchy and debates around it are heated. But send kids to private schools and are of particular, despite profound implications it has for the risk historically advantaged caste/ethnic categories. of conflict, political stability, investment climate and growth in the medium term, current debates The results presented here underscore the urgent on the issue appear to provide very little basis on need to address spatial disparities in access to which to forge a constructive path forward. and quality of education, health and infrastruc- ture services. Many public schools, particularly In this report, we introduce equality of opportu- in rural areas in remote regions lack adequate nity for children as an ideal that could potential facilities and qualified teachers. Similarly, gov- bridge that gap. A society may have disagree- ernment hospitals are poorly equipped, often ments about the sources of past injustices and lacking essential medicine and trained health what to do about them. But everybody can care professionals. Even when these inputs are agree that opportunities for a better life for Ne- available, quality of delivery is low due to weak pali children should not depend on which part accountability mechanisms to local citizens. The of the country they happened to be born in, proliferation of private providers in both health or to what caste /ethnicity group their parents and education in cities and small towns is a re- happened to belong to or what language they flection of an increase in the demand for quality speak or other such characteristics essentially services by an increasingly better off population. determined by the lottery of birth. An inclusive But as the better off flee public services, these society is necessarily also an equal opportunity facilities also become exclusive domains of the society (even though the converse many not be poor, who are often less vocal and assertive in 63 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L their demands for quality which in turn further in Nepal for a “class based” targeting as opposed lowers quality and aggravates inequalities. So the to “caste based” targeting of opportunities. agenda of improving the services for the poor remain a critical piece in the improvement of life Thinking through whether this is appropriate chances of all Nepalis. or feasible for Nepal and operationalizing it has been a challenge. This is not for want of ideas Leveling the Nepal’s decentralization efforts over the last two though. Previous governments have initiated playing field decades have been stunted by conflict and an thinking and some preliminary experimenta- environment of prolonged political limbo. The tion on the idea of poverty scorecards based on for all could country also experimented with devolution of which social assistance could be targeted directly move the debate public schools to local management – in part in to the poor, irrespective of which social or geo- on inclusion response to the difficulties of operating public graphic group they may belong to. The Ministry forward as schools during the time of conflict – but the re- of Poverty Alleviation and Cooperatives was es- sults are mixed. tablished in 2011 to precisely bring this idea to what is a fruition. But the idea appears to have found lit- minimum With the adoption of the new constitution, it tle support in subsequent governments that have criterion for is clear that Nepal is to become a federal state. been occupied with other pressing priorities. fairness and If the outstanding issues around number, names and boundaries of these federal units can be de- In education there are scholarships for children justice for the cided without causing any more damage to the from the Dalit community and for all children left happens collective social harmony, then there are reasons in the Karnali region (the region that lags the also to be the to be optimistic about the possibility of real de- rest of the country in human development and foundation for centralization to finally take root in the coun- other welfare outcomes).To equalize early child- try. Spatial disparities in opportunities can be hood health outcomes, there is a child grant meritocracy for directly addressed by, for example, embedding program which provides an equivalent of $2 per the right spatial differences in welfare and opportunities child per month to mothers with children un- into formulas that determine resource transfers der 5 in the Karnali region and to Dalit mothers from the central to the federal governments. A nationwide. Categorical targeting of a particular meaningful devolution or de-concentration of region or an easily identifiable social group is funds, functions, and functionaries under the perhaps easier to implement – though available federal government structure can help strength- assessments suggest they are not without leak- en incentives for better performance, but it has ages – but it is not entirely adequate from the to be supported by adequate accountability perspective of equity. There is a sizeable number mechanisms to protect against local capture. of poor across Nepal who are non-Dalits and liv- ing in regions other than Karnali. Broadening TARGETING THE POOR, WAY TO MOVE that eligibility base, particularly for these child FORWARD human development linked interventions, to Our results show that the opportunity profiles include the poor everywhere appears to be one for Nepalis today begin to diverge quite dramat- area of policy attention. One example that is al- ically even in childhood. Contrary to percep- ready in place is the secondary school stipend tions about the dominant role of social markers program. This program is designed to help fi- such as ones caste/ethnicity, we see that parental nancially constrained students entering second- wealth and geography (where the child is born) ary school and is a poverty targeted program. are equally if not more important determinants The government is considering ways to expand of life chances. In addition, even within lagging it also to the tertiary level. areas or regions or within the same social group, there is a substantial variation in access to op- When policy conversations turn to expansion of portunities by the economic status of the house- benefits of any kind, a question that inevitably hold. This suggests the need for investments in arises is that of affordability. Does the govern- equalizing opportunities to be based on overall ment have the fiscal space to implement such economic disadvantage instead of social advan- programs? This is a valid concern and a detailed tage. This resonates with commonly heard calls assessment of affordability needs to be done be- 64 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L fore implementing any such program. But two ers by helping villagers understand that women points need to be considered. First, appreciable can be competent leaders.28 The same study also savings may be possible from streamlining exist- finds that exposure to female leaders raised edu- ing programs in the portfolio that is noted to cational aspirations and outcomes for girls in be highly fragmented. Second, the returns on the schools in these villages. Another study finds boosting human capital of children – particu- that women’s chances of election in constituen- larly by improving opportunities for otherwise cies that previously had been reserved for wom- underserved –can be sizeable.26 This return is en leaders was five times higher than in other likely to be even higher for Nepal because its constituencies.29 demographic structure gives it a 25 year win- dow of opportunity to realize the benefits of Reducing Vulnerability demographic dividends.27 A better skilled work- The recent earthquake that Nepal suffered was a force built on the back of aggressive investment violent and a highly visible example of a natural in human capital today will help the country disaster wiping out the lives and livelihoods of maximize these benefits over the next 25 years a large number of Nepalis. As important as it and the investments may be justifiable given the is for the Government to respond to this spe- high returns. cific emergency, this should also serve as an op- portunity to galvanize support for the design of SOME AFFIRMATIVE ACTION MAY BE systems that can respond in a more agile and ef- NEEDED TO REDRESS PAST INEQUALITIES ficient manner, not just to catastrophic disasters Finally while equalizing opportunities for all of similar scale (if need be) but also to smaller children in Nepal is a priority agenda for the but more frequent shocks many Nepalis rou- future, something will have to be done to re- tinely grapple with. These “dull disasters” take a dress the residual effects of past disadvantages heavy toll on poor households by limiting their as well. In this regard, some form of affirmative ability to maintain investments in the human action or positive discrimination policies may be capital of children and depleting their savings required. The new constitution has made pro- and assets. Inadequate protection from these visions for representation of a variety of under- shocks limits the ability of households to take served groups in several of political and public risks and make productivity enhancing invest- offices and there are already provisions for af- ments in both agriculture and non-agricultural firmative action in civil service appointments. activities. But ensuring adherence to these principles in legislation and practice would go a long way. The high degree of vulnerability and transitions Protection Neighboring India has a long history of quotas around the poverty line with suggests the need from shocks in political bodies and educational institutions to strengthen social protection systems, particu- for members of those belonging to Scheduled larly those that insure households against a vari- can also enable Castes/Scheduled Tribes to redress similar his- ety of income shocks. One area of emphasis here households torical inequalities in higher education, employ- should be on designing safety net systems that to take risks ment and politics and these policies have not are able to target and deliver assistance based on and make been free of controversy. The use of quotas in broader, income based notions of vulnerability particular limits the pool of candidates, restricts through vehicles such as means tested condi- productivity free competition and reduces the quality of can- tional cash transfers. More work is required here enhancing didates on average but there is unfortunately no but at the conceptual level, programs designed investments in other way of redressing generational inequalities to specifically to safeguard of investment in the agriculture as in the short run. human capital of children in the face of shocks appear to have strong equity as well as efficiency well as non- Some emerging evidence from neighboring In- arguments. agricultural dia suggests that demonstration effects of these activities. policies could be quite beneficial in leveling Available analysis suggests that existing social playing fields. For example, mandated represen- protection programs have not been very effec- tation of women in local government bodies in tive in protecting the poor: poverty headcount West Bengal reduced biases against female lead- would only increase by a marginal 0.5 percent- 65 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L age points (from 25.2% to 25.7%) in the ab- more to do with tightening of the supply of la- sence of all currently administered social assis- bor than any improvements in labor productiv- tance programs.30 Part of the reason for this low ity. This is partly because in remote areas, espe- impact on poverty is that the programs are not cially in the hills and the mountains, agriculture very generous. But these programs are also not remains overwhelmingly at the level of subsis- explicitly targeted to the poor, and in addition tence. Lack of market integration keeps farmers to suffering leakages and poor administration, autarkic; crop choice and farming methods re- they are also fragmented across several govern- main conservative reflecting high priority placed ment agencies giving rise to rampant duplica- on food security concerns. Lacking year round tion or gaps in benefit coverage. Thus a unified, irrigation, there is near total reliance on rainfall. consolidated safety net system that delivers as- sistance using a common poverty based target- Increased diversification away from traditional ing framework would constitute an important cereal crops would help boost productivity. agenda of moving forward on reducing vulner- Higher value crops such as fruits, nuts, spices, ability in Nepal. herbs, and specific vegetables and livestock prod- ucts have growing demand, not just in the cities Agricultural Finally the high incidence of vulnerability in and emerging towns of Nepal, but also in India. Nepal – when vulnerability is defined strictly in But commercialization also requires better in- commercialization terms of the risk of falling back into poverty – is tegration with markets and value chains. Thus, also requires better also because there appears to be a “productivity strengthening research and extension services, connectivity and ceiling” or a choke point of sorts that is keeping rationalizing input subsidies, improving access integration with Nepalis who have moved out of poverty in this and quality of financial services and improving stage. This is largely due to the fact that a bulk of rural connectivity would all have to be part of markets and the poverty reduction itself has come as a result an integrated strategy to improve productivity value chains. of non-farm diversification in rural areas and in agriculture. As the overseas migration experi- the expansion of service sector jobs in the urban ence deepens, the pool of migrant returnees is areas. But these jobs are predominantly casual also likely to increase over time. These migrants and informal in nature and as such unable to will come not only with resources but also with move people to the next level. Generating more creative ideas and visions formulated through productive jobs within the economy would be rich experiences working abroad. Having the crucial in facilitating continued upward mobil- right policy and investment environment to- ity beyond the vulnerability line. gether with easy access to complementary inputs such as better connectivity and financial services Boosting Productivity will help leverage their creativity and entrepre- The analysis points to three key dimensions of neurial spirit to boost rural growth. boosting productivity that will be important in improving the welfare of the poor and the vul- LEVERAGING THE AGGLOMERATION nerable: boosting productivity in agriculture, POTENTIAL OF CITIES converting the consumer advantage of Nepal’s Nepal has been urbanizing at a rapid pace but cities and emerging towns to a producer ad- the growth benefits of the agglomeration have vantage and finally, making the reforms and in- not fully been realized. This is mainly because vestments necessary to improve the investment most of the increase in urban population in re- climate for a private-sector led growth of the cent years has been due to an increase in urban secondary and tertiary sector of the economy. sprawl around major highways, district head- quarters and cities and a rural to urban migra- IMPROVING PRODUCTIVITY IN THE tion that appears to be driven much more by the AGRICULTURAL SECTOR CRITICAL FOR THE push of the lack of services and amenities the CHRONIC POOR increasingly well off rural population appear to While off farm employment has been the ticket value than the pull of a dynamic urban economy out of poverty, a majority of the rural poor and that is generating good jobs. The negative effects vulnerable remain in agricultural jobs. Agricul- of agglomeration, on the other hand, have been ture has seen very little productivity growth over highly visible as the unplanned and unregulated the last two decades. The growth in wages have growth of the city has not only led to the dete- 66 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L rioration of basic services such as drinking water, is important to expand employment opportuni- Nepal needs to sanitation but also polluted the city’s rivers and ties, particularly wage employment opportuni- generate more air to highly toxic levels. ties in productive sectors outside agriculture. This is the fundamental challenge facing policy- and better As the primary seat of the government, Kath- makers in Nepal today and is also one that will jobs within mandu has enjoyed historical primacy among require a fundamental rethink of the growth the country to cities but over the last two decades, its share of framework that guides development policy. The translate success total urban population has been declining. This fact that Nepal appears to have de-industrialized suggests that there might be an opening to in- rather prematurely raises the question of wheth- in reducing vest in some of the emerging towns to introduce er services that have grown in prominence can poverty into and implement longer terms plans for organized generate the growth in productivity required to success in urban development. Simultaneously, efforts to sustain the level of growth that would be needed generating revive and regenerate the urban Kathmandu to generate the volume of jobs that would need valley to transform its comparative advantages, to be created to absorb not only the new labor a large and particularly in tourism and labor intensive man- market entrants but, potentially also absorb the vibrant middle ufacturing in handicrafts into competitive ad- stock of those that are currently out of the coun- class. vantage would also be a key part of the strategy. try. Available evidence suggests that growth in services may have maxed out mainly because the Nepal’s current industrial policy provides tax key driver (remittances) is already at a high level incentives and subsidy for investment in the de- and is unlikely to go up further. Additionally, velopment of the plant for firms to locate their unlike India’s IT sector for example, the sub- productive activities in lagging regions. These sectors within services that have shown the most “spatially targeted” policies represent the gov- potential for growth have been non-tradeables. ernment’s deliberate efforts to de-concentrate productive activities. But locating production Be it industries or services, stabilizing the po- activities in remoter locations with lower popu- litical environment, carrying out the necessary lation densities is seldom effective in propping reforms to improve the investment climate and up lagging regions or generating growth. In fact, alleviating the crippling infrastructure deficit are if enforced, such policies could be detrimental critical entry points into the agenda on jobs. for the growth of the industrial sector. Instead, from the perspective of growth, a better policy Finally, it must also be recognized that there is would be to invest aggressively in enhancing no level of growth Nepal can realistically generate connectivity of the remote regions to the emerg- in the medium term that will generate jobs at a ing growth centers or to invest in the develop- fast enough pace to completely overturn the flow ment of these growth centers themselves. of migrant workers. A more realistic outlook is that a steady state level of overseas migration will MORE AND BETTER JOBS IN NEPAL remain a critical dimension of Nepal’s jobs envi- Off farm diversification may have been the ronment in the medium to long run. For this rea- ticket out of poverty but Nepal needs to gener- son, building a better understanding of not just ate more and better jobs within Nepal to see its the migration process in general but also of what middle class grow. In order to ensure continued can be done to make the process more efficient improvements in the wellbeing of those at the and less costly so that successive cohorts of Nepali bottom of the distribution and in particular to migrants have better information of opportuni- ensure that the success in lifting people out of ties abroad, can become more productive and can poverty is taken the additional mile to move have access to better quality jobs abroad appears them robustly above the vulnerable category, it an important part of the jobs agenda. 67 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 68 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L REFERENCES Azevedo, João Pedro, Gabriela Inchauste, Jaime Case, A., and C. Paxson. 2006. “Stature and Status: Saavedra, and Hernan Winkler. 2013. “Is Labor Height, Ability, and Labor Market Outcomes.” Income Responsible for Poverty Reduction? A NBER Working Paper 12466, National Bureau of Decomposition Approach.” World Bank Policy Economic Research, Cambridge, MA. Research Working Paper . No. 6414. April 1. Chetty, R., J. Friedman, N. Hilger, E. Saez, Barros, R, F. Ferreira, J. Molinas Vega and D. Schanzenbach, and D. Yagan. 2010. “How J. Saavedra (2009). “Measuring Inequality Does Your Kindergarten Classroom Affect of Opportunities in Latin American and the Your Earnings? Evidence from Project STAR.” Caribbean.” The World Bank. NBER Working Paper 16381, National Bureau of Economic Research, Cambridge, MA Barros, R, J. Molinas Vega and J. Saavedra (2010). “Measuring Progress Toward Basic Currie, J., and D. Thomas. 1999. “Does Head Opportunities for All”. Brazilian Review of Start Help Hispanic Children?” Journal of Econometrics, 30 (2). Public Economics 74 (2): 235–62. Banerjee, Abhijeet and Esther Duflo (2008). Deaton, A., & Zaidi, S. (2002). Guidelines for “What is middle class about the Middle Classes constructing consumption aggregates for welfare around the World”. Journal of Economic analysis (No. 135). Perspectives 22(2):3-28 Easterly, William (2001). “The Middle Class Behrman, Jere R., Susan W. Parker, and Petra Consensu and Economic Development”, E. Todd. 2005. “Long-Term Impacts of the Journal of Economic Growth. 6(4): 317-35 Oportunidades Conditional Cash Transfer Program on Rural Youth in Mexico.” IAI Ferreira, Francisco H. G., Julian Messina, Jamele Discussion Papers 122, Ibero-America Institute Rigolini, Luis-Felipe Lopez-Calva, Maria Ana for Economic Research, Berlin, Germany. Lugo and Renos Vakis. (2013). “Economic Mobility and the Rise of the Latin American Bhavnani , R. R. 2009. “Do electoral quotas Middle Class” World Bank, Washington DC work after they are withdrawn? Evidence from a natural experiment in India” American Haughton, J., & Khandker, S. (2009). Political Science Review. 103(1) 23-35 Handbook on Poverty and Inequality. World Bank Publications. Birdsall, N. (2010). “The (Indispensable) Middle Clsas in Developing Countries; or, Heckman, James J. 2006. “Skill Formation and the Rich and the Rest, Not the Poor and the the Economics of Investing in Disadvantaged Rest”, Working Paper #207 Center for Global Children.” Science 312 (5782): 1900–02. Development, Washington DC Hoddinott, John, John A. Maluccio, Jere Birdsall, N., F. Ferreira, L.F. Lopez-Calva R. Behrman, Rafael Flores, and Reynaldo and J. Rigolini (2011). “The Middle Class in Martorell. 2008. “Effect of a Nutrition Developing Countries: Concept, Measurement Intervention during Early Childhood on and Recent Trends”. Mimeo. The World Bank, Economic Productivity in Guatemalan Adults.” Washington DC The Lancet 371 (9610): 411–16. http://dx.doi. org/10.1016/S0140-6736(08)60205-6. 69 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Lopez-Calva, L.F. and E. Ortiz-Juarez (2011). Shrestha, Mahesh. (2016a). “Get Rich or Die “A Vulnerability Approach to the Definiiton Tryin’: Perceived earnings, perceived mortality of the Middle Class”. Policy Research Working rate and the value of a statistical life of potential Paper 5902. The World Bank, Washington DC. work-migrants from Nepal”. Job Market Paper, Department of Economics, Massachusettes Maluccio, J. A., Hoddinott, J., Behrman, J. Institute of Technology. R., Martorell, R. Quisumbing, A. R. and A. D. Stein (2009), “The impact of improving Shrestha, Mahesh. (2016). “Death scares: How nutrition during early childhood on education potential work migrants infer mortality rates from among Guatemalan adults” Economic Journal, migrant deaths”. Job Market Paper, Department of 119(537): 734-763 Economics, Massachusettes Institute of Technology. Mitra, S. (2014). “Synergies Among Monetary, Tiwari, Sailesh, Hanan Jacoby and Emmanuel Multidimensional and Subjective Poverty: Skoufias (2016). “Monsoon Babies: Rainfall Evidence from Nepal.” Social Indicators Shocks and Child Nutrition in Nepal”, Research. http://doi.org/10.1007/s11205-014- fortcoming in the Economic Development and 0831-3 Cultural Change. Muzzini, E., & Aparicio, G. (2013). World Bank. (2006). “Nepal Resilience “Urban Growth and Spatial Transition in Amidst Conflict An Assessment of Poverty Nepal: An Initial Assessment. Directions in Nepal, 1995-96 and 2003-04” Retrieved in Development”.World Bank Directions in from http://documents.worldbank.org/ Development http://doi.org/10.1596/978-0- curated/en/2006/06/6919674/nepal- 8213-9659-9 resilience-amidst-conflict-assessment-poverty- nepal-1995-96-2003-04 Newman, John L., João Pedro Azevedo (2013). "Setting reasonable performance targets for World Bank (2012). “More and Better Jobs in public service delivery," Policy Research Working South Asia”. World Bank South Asia Development Paper Series 6385. The World Bank Matters, World Bank Washington, D.C. Oxford Poverty and Human Development World Bank. (2013). “Nepal Bigger Strides Initiative (OPHI). (2013). Country Briefing : in Post-Conflict Poverty Reduction Poverty Nepal Multidimensional Poverty Index (MPI) At Diagnostics, 2010-11”. Washington, D.C. a Glance. Retrieved from http://www.ophi.org. uk/wp-content/uploads/Nepal-2013.pdf World Bank (2014). “A Vision for Nepal: Policy Notes for the New Government”, World Bank, Ravallion, Martin (2010). “The Developing Washington DC. World’s Bulging (but Vulnerable) Middle Class”. World Development. 38(4):445-454 World Bank. (2015). “A Measured Approach to Ending Poverty and Boosting Shared Rodrik, Dani (2015). “Premature Prosperity: Concepts, Data, and the Twin Deindustrialization” NBER Working Paper No. Goals”. Policy Research Report. http://doi. 20935, NBER Boston org/10.1596/978-1-4648-0361-1 Sen, Amartya. (1983). “Poor, Relatively World Bank. (2016a). The World Development Speaking”. Oxford Economic Papers. 35(2): Indicators. Retrieved from http://data. 153-169 worldbank.org/data-catalog/world- development-indicators Shilpi, Forhad, Prem Sangraula and Yue Li (2014). “Voting With Their Feed: Access to World Bank (2016b). “Remittances at Risk: Infrastructure and Migration in Nepal”. Policy Nepal Development Report”. World Bank, Research Working Paper Series #7047. The World Kathmandu Nepal Bank 70 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L END NOTES 1 World Bank (2014), “A Vision for Nepal: Policy verted to a permanent status. Permanent employees Notes for the Government” are eligible for participation and protection by unions and among other benefits, they enjoy significant secu- 2 The revised poverty line is Rs.19,261 per person per rity of tenure. Firms respond to this by rationing their year in 2010/11 prices (World Bank, 2013) pool of regular employees and relying on contract la- 3 Based on benchmarking methodology described in bor. This has hiring-firing rigidity has been cited as Newman, John L., João Pedro Azevedo (2013) "Set- one of the primary constraints to firm growth. ting reasonable performance targets for public ser- 16 Dang and Lanjouw (2013) vice delivery," Policy Research Working Paper Series 6385, The World Bank (http://ideas.repec.org/p/wbk/ 17 Tiwari et al (2016) wbrwps/6385.html). 18 Lopez-Calva and Juarez-Ortiz (2011) 4 There are several estimates of Multidimensional Pov- 19 A number of other efforts have been made to de- erty available in Nepal from various data sources. For fine the middle class globally. Birdsall (2010) and example, see OPHI (2013) and Mitra (2014). Birdsall et al (2011) proposed a relative concept of Based on data from various rounds of Population Census. 5 the middle class defined as those between 75% and 125% of any society’s median income. The notion Personal remittances received as percent of GDP is 8.7 per- 6 that the definition of the middle class should be dif- cent for Bangladesh and 3.4 percent for India as of 2014. ferentiated between developing countries and other, 7 Despite this, Nepal remains the least urbanized coun- wealthier nations has also be noted quite generally. try in the region. Ravallion (2010) for example proposed consumption level between $2 and $13 as middle class for develop- 8 Inequality estimates based on the 2003/04 Nepal Liv- ing countries while the “Western middle class” line is ing Standard Survey were significantly higher than anybody above the US poverty line. But a lot of these those in 1995/96 and 2010/11 but there is significant approaches end up lumping the poor and the vulner- concern for measurement errors and “the up and down able into the middle class category. swings are of such a magnitude that they are nearly non-credible” (World Bank, 2015). For this reason and 20 Ferreira et al 2013, Easterly 2001 also for the sake of brevity, this Chapter focuses on in- 21 Ferreira et al (2013); Easterly (2001) equality estimates in 1995/96 and 2010/11. 22 Shipli et al (2014) 9 The so called “Patidar Reservation Agitation” in India around July-August 2015 is an example of an instance 23 This is a fundamental difference between the new eco- when an otherwise privileged caste group seeks inclu- nomic geography literature and the classic models of sion into the excluded group category in order to be rural-urban migration due to Todaro. Both start from able to access mandated quotas in higher education the same place: surplus labor in rural agriculture and and government jobs. rural to urban migration driven by the pull of higher productivity manufacturing jobs in the cities. But in 10 See for example Alvaredo and Piketty (2014) on an Todaro’s world surplus growth in manufacturing would application of this methodology to the Middle East draw rural workers to cities to such an extent that that and North Africa Region. would lead to unemployment in urban areas. The new 11 There may be some ambiguity about when effort ac- economic geography literature – which is influenced by tually becomes relevant for a child, which is likely to the then emerging endogenous growth literature which depend on the social, cultural and ethical norms that recognizes the possibility of increasing returns – suggests vary across societies. A more pragmatic approach is to that unemployment need not rise in order to absorb the consider effort as being irrelevant for children below age 24 Muzzini and Aparaicio (2013) sixteen, which would dilute the concern about possible correlations between effort and circumstances. 25 See Rodrik (2015) for a more detailed exposition of this idea more generally. 12 Currie and Thomas (1999) and Case and Paxson (2006). 26 Heckman (2012) 13 Chetty et al.(2010) 27 World Bank (2012) 14 Alderman et al (2006) and Hoddinott et al (2008) 28 Beaman et al (2009) 15 Nepal’s labor law stipulates that any employee ap- pointed to a job over a continuous period of 240 days 29 Bhavani (2009) by firms employing more than 12 employees be con- 30 World Bank (2013) 71 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 72 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L ANNEX ANNEX 1: Human Opportunity Index (HOI) Methodology A SIMPLE EXAMPLE portant caveat of this measure is that it is sensitive Consider two societies A and B in which half the to the set of circumstances chosen for analysis. But population lives in rural areas and the other half this is mitigated by an additional property that is in urban areas. Now consider a basic opportunity highly desirable given that it is often impossible to such as access to primary education. Say, 50 per- identify all relevant circumstances for any popula- cent of all children go to school in both the soci- tion and opportunity: the HOI will not be higher eties. Looking at the overall coverage, both these if more circumstances are added to the existing set societies will appear similarly placed. But, suppose of circumstances in the analysis. This implies that we also know that in society A, no rural child at- the computed inequality serves as a lower bound to tends a school; while in society B, 50 percent of the “actual” inequality where all circumstances of both rural and urban children attend school. The interest could be included in the analysis. HOI discounts the coverage rate of 50 percent by imposing a “penalty” when access is more unequal ESTIMATING THE HOI FROM inequality serves as a lower bound to the “actual” inequality where all HOUSEHOLD based on circumstances such as location. The im- SURVEY DATA included in the analysis. posed “penalty” can be interpreted as the share of To construct the HOI, we need to obtain the the total number of opportunities that need to conditional probabilities of access to opportuni- Estimating the HOI from Household Survey Data be re-distributed to ensure equitable access based ties for each child based on To theirconstruct circumstances. the HOI, we need obtain the conditional probabilities to be inequality serves as a lower bound to the “actual” inequality where all circumstances of interest could on the equality of opportunity principle. included in the analysis. Going In inequalityorder to serves do asso,a one lower can child bound estimate based to the on atheir “actual”logistic circumstances. inequality where Inallorder to do so,ofone circumstances can could interest estim back to inequality the previous serves example, as a lower in society boundA, to this the “actual” model, included inequality linear in the analysis. in the where parameters β, of all parameters circumstances where where interestthethe event couldI be corresponds to accessing the opportuni included will constitute in the analysis. “reallocating” 25 percent of total event I corresponds to accessing the set of the circumstances, opportunity(e.g. gender of the child, education and gender inequality serves as a lower Estimating the bound HOI from to Household the “actual” inequality Survey Data where allWe fit the logistic circumstances regression of interest could using be survey data: enrollments from urban included children in the to rural children. (e.g. access to clean water), and x the set of cir- Toanalysis. Therefore, the inequality penalty would be construct the HOI, we needEstimating 25 percent to obtain the the conditional probabilities HOI from Household Survey Data of access to opportunities for ⎛ P [I = 1 X = (x1 ,..., x m )] ⎞ m each serves child as a based lower theirand on bound circumstances. to the “actual” cumstances, In order inequality (e.g. to gender so, do where oneall of can the estimate child, circumstances education a logistic model, linear interest ofconditional could inbe Ln ⎜ the ⎟=∑ xk β − P [I = 1aX logistic x m )] ⎟ Estimating the HOI from Household Survey ToData construct the HOI, we need to obtain the probabilities ⎜1 of access to opportunities for the HOI, which is the included incoverage parameters the analysis. minus the pen- β, where the event I and corresponds child gender based to of on accessing the head their the opportunity circumstances.of the household, (e.g. access toetc). In order to clean water), and do so, one can ⎝x estimate = (x 1 ,..., model, ⎠ linear k =1 in construct To equal the HOI, the set we of needcircumstances, to obtain (e.g. gender of the child, education and gender of the head of the household, etc). alty, would Estimating 25 percent. the For HOI fitsociety from BHousehold in our the conditional Survey We parameters fit Data the probabilities logistic β , where the regression of access event using I correspondsto survey opportunities to data: accessing for the each opportunity (e.g. access to clean water), an child based To on their We construct circumstances. the the logistic HOI, we regression need In to order obtain using to the the survey do set so,data: conditional of one can estimate probabilities circumstances, (e.g. Where of aaccess gender xkto logisticof denotes the model, opportunities child, the linear education row for in vector eachthe gender and of variables representing the k-d of the head of the household, there is no example,parameters inequality based on location β, where child based the eventcircumstances. on their I corresponds Into accessing order We⎛ to fit [ PtheIdo =1 theX =opportunity so, logistic one can ,..., (x1regression ] x m )estimate x ⎞ (e.g. using= m (ax ,..., access 1logistic survey x to data: m ) and clean water), model, β ′ = linear 1 ( β in ,..., and the βxm ) is a corresponding column vector and the penalty the set ofEstimating is parameters zero. This β circumstances, the HOI implies , where (e.g. from that thegender Household the HOI event of I Survey the child, education corresponds Ln ⎜ to Data − P [I = 1 X ⎜ 1 accessing andthe ⎟ 1 ,..., x m )] ⎠ ofk = = (x gender opportunity of = ⎟ the ∑(e.g. this x β head k ofto k access logistic the cleanhousehold, regression water), one and etc). x obtains estimates of the parameters {β k } We fitor is 50 percent, To the construct the equal set logistic to of the HOI,using circumstances, regression the coverage. weTherefore, need (e.g. surveygender to data: obtainof the the ⎝ conditional child, education probabilities and gender of of 1 access the ⎛head Pto[ =opportunities I of 1 X= the (x1 ,..., x m household, for ] )etc).each ⎞ m child based on their We fit the logistic regression Where circumstances. x denotes using survey data: the In row order vector toof do so, variables one can representing estimatedenotes the Ln k a ⎜ the sample size. ⎜ 1 − P [I = 1 X = (x ,..., x )] ⎟ logistic -dimension model, of Given linear circumstances. in ⎟the the = ∑estimated Thus, x k β k coefficients, one can obta β[ )] the society B is more equal parameters than β, society where A the k based ,...,event on the I corresponds Where is ax k denotes the row vector his/her(e.g.⎝of of variables predicted rep- probability 1 the of m access⎠ xk =1to the opportunity in consideration x = (x1 x m ) and ⎛ P ′I= = (β11X = ,..., β (to x1 accessing ) ,..., xcorresponding ⎞ opportunity m column vector access to clean parameters. From water), and estimation ⎜ aver-⎛of P [Iobtains m m ⎟) ]= ∑ β k of ⎞ gender ( ) m k } to be denoted by { equality of opportunity the set ofcriteria, even though Lngender = 1 X = (x resenting ,..., the xm k⎟-dimension x ofcircumstances. of theThus, ˆ } where P[ m )] circumstances, (e.g. the child, education and the head household, etc). We fit the logistic of this logistic regression ⎜ 1 −Ln regression ⎝ using survey ⎜I one ⎜ =1 XWhere [ data: = (x 1estimates x1k ,..., xdenotes of ⎠ the ] ⎟ the ⎟ = ∑ krow parameters x k β kvector{β of variables representing β k ,n then k-dimension Exp ˆ of circumstances. xi β ˆ n T i ,n = age enrollment rate is the same in both − = = = the societies. k 1 1 P I 1 X ( x ,..., x ) vector of p Where xkWhere denotes denotes his/her the row sample size. ⎝ predicted vector probability Given the of ⎛variables [of access x= (x1 ,..., estimated to logistic representing 1x m m ⎠ βk ) and coefficients, the opportunity )] the ′= in k =1 (β 1 one can -dimension ,...,obtain consideration: β m ) is is of fora correspond- a each corresponding individual in column the sample parameters. Thus, {β k } to be denoted by ( From 1 + Exp x i β ) ˆ { the estima β k ,n } whe nˆ xk denotes the row vector P of I= ing 1ofXthis variables column = (x1 ,..., x vector representing ⎞of mregression the parameters. m kone obtains -dimension From of circumstances. the estima- estimates of the parameters circumstances. Thus, x = ( x THREE KEY PROPERTIES1 ,..., x x= ) and β ′ = ( β m (x1 ,..., x m ) and 1β ′ = (β ,..., β m ) Ln is 1 ,..., β⎜ a⎜ m1 )− isPa [ corresponding I tion corresponding =denotes 1 X of= (this the x columnlogistic sample ,..., column x )] ⎟ Exp = ( vector ⎟size. x ∑ ˆ regression βn vector ) x of Givenk β The k one overall parameters. of parameters. obtains the estimated coverage From the estimates From rate, coefficients, estimation the estimation C, one can obtain for each individual inthe the D-Index, the penalty, P, and theHO sam ⎝ { } 1p ˆ i ,n =m ⎠ k =1 i First, theof HOI this is sensitive logistic of thisregression to scale logistic one– ifobtains regression access one obtains im- estimates his/her of estimatesof the the predicted ofparameters the parameters 1 + Exp {β (xki{β probability } ˆ βnk )} to toprobability toof access bebe denoted denotedto denotedby ˆ p theand by { sampling opportunity by β ˆ ˆ βkk,,n n } weights, in where where nn w consideration: : proves for all groups denotes Where by, the sample denotes xk denotes say, the a factor size. sample The the row Given overallsize. β ′ = (ˆ of the k vector Given coverage β 1 ,..., (addi- estimated β the rate, of variables representing the k-dimension of circumstances. C, coefficients, estimated thewhere D-Index, n denotes coefficients, one the can one penalty, theobtain can P,sample and for obtain the forsize. each HOI each Given individual are individual estimatedthe in esti- in using the Exp (nx i βn the ) Thus, ˆ predicted sample sample 1 n his/her x =his/her (x1 ,..., predicted predicted x m ) and probability probability tively or multiplicatively), then the HOI changes probability of access p and of access to ) is a to m the weights, sampling corresponding the opportunity opportunity w : in column in consideration: mated coefficients, one can obtain for each indi- vector consideration: of parameters. ˆ p From i ,n = C the = ∑ estimation ( w p iˆˆi ,n ; D = ∑ w i p ) ˆi ,n − C ; P = C * D ; an of this logistic regression one obtains estimates of the nˆ(x isample β n ) rate, parameters { β } to be denoted by { 1+ β ˆ i=Exp } x β i n 1 where n 2C i =1 by the same factor k. Second, it rewards Pareto n vidual ∑ The p Exp ˆcoefficients,( inExp = 1x i βcoverage overall ∑ the ) ˆ his/her k C, *the predicted D-Index, probability the− Ppenalty, k , n and the estimation P, sample HOI are estimated model isusing that thethe predi improvement– denotes if coverage the sample size. Given the rate improves for one his/her predicted probability of access i to C= estimated pˆw ip i ,n ˆ=i ,n ; of =1 the opportunity i ,D probability n= access 1 + Exp ix 21 C+ ( toExpˆ p =1 β w i one n the i n ˆ( ) and in x pˆi ˆ i β )can obtain ,n − C ; P = C opportunity sampling consideration: n An Dfor in weights, interaction and each ;important w : HOIindividual consideration: terms =C caveat between toin the logistic circumstances (e.g. between parental education list circumstance-group without decreasing rate, coverage rates for The overall the remaining The overall coverage coverage groups, An important rate, the C, HOI the C, theto caveat D-Index, will D-Index, the logistic : the penalty, rise.wcircumstances the estimation penalty, P, Exp (and P, between ˆand model xi β ) parental nthe the is that HOI education C =∑ HOI circumstances n the are arelistestimated of regressors ˆestimated we nusing using have 1 opportunities doesthe(all the − predicted not of include predicted which anyare dummy variables), limited sam probability probability ˆ andalways ˆinteraction p sampling and sampling terms weights, between weights, w : (all of which are pˆ i ,n = (e.g. Exp(x ivariables), ˆ ) countries w ip and i ,n ; D = and ∑ location). w i Given ˆ p the for number which of these regressions have to be run, i ,n C ; P = C * D ; and HOI = C − P Third, the measure p will circumstances improve we if have access 1+ dummy β n limited sample sizes, i =1 intractable problems 2C i =1and the in largeat least number some of of the cases. The interaction te countries and opportunitiesn for 1which n these regressions have to translating be run, including the exactinteractionsdefinition would of lead to D-Index to the logistic regression m changes in such The a way overall that the coverage more intractablen rate, = ∑w vulnerable C problems C, theip ˆiin ,n 1 D= ;at D-Index, n The least An∑w the some overall pˆi ,the important penalty, i of n −C coverage P,;caveat cases. Pand= * C the The toD rate, ;the andlogistic HOI interaction C, HOI terms.are the = − estimation estimated terms If C the D-Index, P thus are using interactions model the omitted, the were is thatthough predicted even the list of regressors does not include included, it would result in a higher D-Ind with coverage groups (groupsprobability pˆ andC = rates terms. ∑ translating sampling lower If w ˆ the p i exact ; than the i iweights, = ,n 1 the interactions D = w: ∑ penalty, definition 2C were 2wCofinteraction i included, p =ˆ D-Index 1 i i ,n circumstances − C ; terms to it would we P = the C P, and the HOI are result between logistic * D have ; and in a (all circumstances regression higher HOI happen of estimated D-Index which = model C − P more if (and (e.g. would are dummy between using require circumstances lower HOI), the parental variables), including just as were education these it would limited and location). Given the numbe added. This in turn implies that the sample sizes, and the large numbe An important overall coverage rate) have higher caveat i = 1 access.to the logistic An im- estimation i = 1 model is that the list of regressors does not include any happen if more n circumstances were predicted n added. and countries probability This in turn implies opportunities and and that for opportunities sampling the which estimated these weights, is the D-Index regressions wlower : have for allbound countriesof inequality of opportunities (and the to be run, including interactions would lea interaction terms between circumstances 1(e.g. between parental education and for location). given asome set Given ofHOI the number circumstances. of An important interaction caveat circumstances terms to opportunities and we for between a theC= have given ∑ logistic set(allw of circumstances ip of ˆiestimation is the lower bound ,n ; D = which are circumstances. (e.g. 2 ∑ model of ˆi ,is winequality i p intractable C dummy between parental translating n − that Cof variables), the ;P = Clist the opportunities problems * Dat in limited education exact of ; andregressors (and least sample and definition HOI location). of the = sizes, Cof and D-Index −does estimated Pthecases. the Given to not the isinclude The large number the the upper number logistic any bound) terms are thus omitted, even tho interaction of of regression model would require73 including t i =1 i =1 countries and opportunities for which these regressions have to be run, including interactions would lead to circumstances we intractablehave (all problems ofin which at least are somedummy of the variables), terms. If cases. the The limited interactionssample interaction were terms sizes, included, are and thusit the wouldlarge omitted,resultnumber even in a though of higher D-Index (and lower HOI), just as it w An important caveat to the logistic estimation happen model is that the list ofwereregressors does in not include any countries and opportunities translating interaction exactfor the between terms which of definition these circumstances D-Index to the if regressions (e.g. between more have logistictocircumstances parental be run, including regression education Hypothetical model and would added. This interactions location). Example require Given turn would the to including implies lead Illustrate these number tothat the estimated D-Index for all coun of Shapley Decomposition Hypothetical Example to Illustrate Shapley Decomposition ⎜ 1 − P [I = 1 X = (x ,..., x )] ⎟ ∑ ⎝ 1 m ⎠ k =1 Where xk denotes the row vector of variables representing the k-dimension of circumstances. Thus, x = (x1 ,..., x m ) and β ′ = (β 1 ,..., β m ) is a corresponding column vector of parameters. From the estimation of this logistic regression one obtains estimates of the parameters {β } to be denoted by β M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L k ˆ where n { k ,n } denotes the sample size. Given the estimated coefficients, one can obtain for each individual in the sample his/her predicted probability of access to the opportunity in consideration: ˆ i ,n = p ( ˆ Exp x i β n ) ( ˆ 1 + Exp x i β n ) The overall coverage rate, C, the D-Index, the penalty, P, and the HOI are estimated using the predicted ˆ and sampling weights, w: probability p n 1 n a total surplus generated by the coalition of all play- C= ˆi ,n ; D = ∑ w i p ∑wi p 2C ˆi ,n − C ; P = C * D ; and HOI = C − P ers using the following rule. In the coalition each i =1 i =1 An important caveat to the logistic estimation model is that the list of regressors does not player include demands any their marginal contribution to the important (e.g. An circumstances interaction terms between caveatbetween the logistic toparental education estimation and location). Given the number of circumstances we have (all of iswhich surplus as a fair compensation, where the player’s model thatare the list ofvariables), dummy regressorslimited does sample sizes, and the large number of not include countries and opportunities for which these regressions have to be run, including interactions contribution would lead to is calculated as the average addition to intractable problems any interaction in at least some of the terms between cases. The interactioncircumstances terms are thus omitted, even though translating the exact definition of D-Index to the logistic regression model would require including the surplus these over all possible different permutations (e.g. between parental education and location). terms. If the interactions were included, it would result in a higher D-Index (and lower HOI), just in as which it would the coalition can be formed. We apply the Given the happen if more circumstances were number added. This in circumstances of turn implies that thewe have (all estimated D-Index for all countries same and opportunities is the lower bound of inequality of opportunities (and the estimated HOI is the upper bound) concept to the decomposition of the D-index, of which are dummy variables), limited sample for a given set of circumstances. with the circumstances being analogous to the play- sizes, and the large number of countries and op- ers, the total D-Index is analogous to the surplus, portunities for which these regressions have to be Hypothetical Example to Illustrate Shapley Decomposition and the different combinations of circumstances are run,of The D-Index or inequality including opportunity interactions index measures the would lead inequality to intrac- between groups. For ato the coalitions in a cooperative game. circumstance equivalent tablethe given set of circumstances, problems in at “contribution” of least some circumstance an additional of the cases. The to the index can be interpreted in the value of theterms as the marginal change interaction D-index are adding afterthus the “new” even circumstance. In other Circumstances that add words, the contribution of a circumstance omitted, though more to the D-Index are then considered as contributing to (or explaining) a larger share of to thethe D-index is the average addition to the value of inequality translating between groups. However, the exact the estimating marginal definition contributionof of D-Index to theis complicated. This is each circumstance the D-index over all possible different permutations because the change in the value regression logistic of D-index as model would require a “new” circumstance is addedinclud- depends on the existing set of circumstances to which the circumstance is added, which in turn implies that the contributions in which of circumstances can be combined. ing these terms. If the interactions were included, it would result in a higher D-Index (and lower Suppose the dissimilarity index (D) is estimated HOI), just as it would happen if more circum- using three circumstances: A, B, and C. Let Dj stances were added. This in turn implies that the circumstances put,denote would change depending on the sequence in which different circumstances are added. Simply of each circumstance of the contribution the contribution unique. Moreover, A circumstance is not the to the of all circumstances contributions estimated D-Index for all countries and oppor- estimated this way may not add up to the full D-index (based on all circumstances), which is not appealing. D-Index, which is the same as the marginal im- tunities is the lower bound of inequality of op- The decomposition based on the Shapley value, which is a solution concept in cooperative game theory, is a pact method thatof a unique,circumstance adding offers intuitive solution to the A above on the D-Index. problem. The Shapley value assigns a unique portunities (and the estimated HOI is the upper distribution among the players of a total surplus generated by the coalition of all players using the following Since the circumstances are correlated rule. In the coalition each player demands their marginal contribution to the with each surplus as a fair compensation, bound) for a given set of circumstances. other, where the the player’s marginal contribution impact is calculated will as the differ average additiondepending to the surplus over all possible different permutations in which the coalition can be formed. We apply the same concept to the decomposition of the on with D-index, which set of pre-existing the circumstances being analogous to the circumstances A is players, the total D-Index is analogous to the surplus, Hypothetical Example to and the different combinations of circumstances are equivalent to the coalitions in a cooperative game. In added to. That is the it will be different whether A Illustrate Shapley Decomposition other words, the contribution of a circumstance to the D-index is the average addition to the value of the D- added isover index to the all possible setpermutations different {B,C}, {B} {C}. Therefore, or circumstances in which to can be combined. The D-Index or inequality of opportunity index mea- Suppose the dissimilarity index (D) is estimated using three circumstances: A, B, and C. Let D denote the measure the “contribution” of A to the D-Index, � sures the inequality between circumstance groups. contribution of circumstance A to the D-Index, which is the same as the marginal impact of adding the Shapley decomposition takes the average of For a given set of circumstances, the “contribution” circumstance A on the D-Index. Since the circumstances are correlated with each other, the marginal impact will differ depending on which set of pre-existing circumstances A is added to. That is the it will be different all marginal impacts when A is added to all pos- of an additional circumstance to the index can be whether A is added to the set {B,C}, {B} or {C}. Therefore, to measure the “contribution” of A to the D- sible Index, subsets the Shapley of the set decomposition takes {B,C}. the average This is given of all marginal by: impacts when A is added to all possible interpreted as the marginal change in the value of the subsets of the set {B,C}. This is given by: D-index after adding the “new” circumstance. Cir- 2 1 1 2 �� = [�(�� �� �) − �(�� �)] + [�(�� �) − �(�)] + [�(�� �) − �(�)] + [�(�) − 0)] 6 6 6 6 cumstances that add more to the D-Index are then Why are the weights for the different components of (A) the way they are? The idea is that we are taking the considered as contributing to (or explaining) a larger average of all possible paths of addition of circumstances. In the figure below, the differences are represented Why are the weights for the different compo- share of the inequality between groups. However, es- by blue lines labeled 1-4. There are six possible paths to go from (0) to D(A,B,C). It is easy to see that two of nents these paths use of blue the (A) the wayas line marked they (4), two are? The use the blue idea is that line marked (1) andwe one path uses the blue lines timating the marginal contribution of each circum- (2) and (3). The weights in the different components of (A) are allocated accordingly. are taking the average of all possible paths of ad- stance is complicated. This is because the change dition of circumstances. In the figure below, the in the value of D-index as a “new” circumstance is differences are represented by blue lines labeled added depends on the existing set of circumstances to 1-4. There are six possible paths to go from (0) which the circumstance is added, which in turn im- to D(A,B,C). It is easy to see that two of these plies that the contributions of circumstances would paths use the blue line marked as (4), two use change depending on the sequence in which different the blue line marked (1) and one path uses the circumstances are added. Simply put, the contribu- blue lines (2) and (3). The weights in the differ- tion of each circumstance is not unique. Moreover, ent components of (A) are allocated accordingly. the contributions of all circumstances estimated this way may not add up to the full D-index (based on all D(A,B,C) 1 circumstances), which is not appealing. D(A,B) D(B,C) The decomposition based on the Shapley value, 2 D(A,C) which is a solution concept in cooperative game theory, is a method that offers a unique, intuitive D(B) 3 D(A) D(C) solution to the above problem. The Shapley value assigns a unique distribution among the players of 4 0 74 the amount of information available on the household, the richer the analysis that can be done. These “panel three minimum, minimum, it is necessary it to is necessary be able to to be able observe athe to observe same the household in athouse same least proper study of economic datasets” mobility as they are calledthat is rounds being are fairly to of hard proposed come NLSS has by in that many have fairly been demanding developing completed dataIn countries. have requirements. the case anAt of Nepal, integrated the the three roundspanel component, of NLSS that have i.e., been subset completed of an have the integrated panel three rounds of NLSS sampled that have been households completed have have been an integrated tracked panel over time, a butthe amount due the of tosampled information sampleamount householdsof attrition, information available have has tracked it been on available the household, been found on over time, thethe that household, richer due to sample attri but the the the ric analysis inimum, it is necessary to be able to observe the same household in at least twocomponent, time periodsi.e., andsubset the of the richer A sampled households have been tracked over time, but due to sample attrition, it has been found that the proper study datasets” as of datasets” they survivingeconomic are as calledthey households mobility are are fairly are nocalledthat hard longer are tois being fairly come representativehard by proposed in of to many thecome nationhas by developing as afairly in many whole. de cou e amount of information available on surviving households the household, are the the richer longer representative no analysis that can be done. ofThese the nation “panelas a whole. surviving households are no longer representative of the nation as a whole. M O V I N G U P minimum, three rounds it the is Asthreeof necessary NLSS rounds that have of ableNLSS been tothat completed been have have methodology an completed integrated have inpanel anatintc the nameto be observe the same household le THE LA DDER : P O V E R T Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E P A L tasets” as they are called are fairly hard to come by in many developing countries. In the case of Nepal, suggests, the proposed overcomes these sh the sampled amount of households sampled information have householdsbeen available tracked have on over been the time, tracked household, but due overthe to time,sample richer but the attriti due anal t ree rounds of AsNLSS namehave the that suggests, As beenthe the name completed proposed have suggests, an integrated methodology thepanel overcomes proposed component, these methodology shortcomings i.e., a subset and overcomes builds of the on an imputation these shortcomings and builds on an based methodology to construct a “synthetic panel” that makes surviving households surviving were observed are no households in two longer time representative are no periods. longer The of the nation representative approach as of a the whole. nation imputation based mpled households have been tracked methodology to imputationconstruct over time,based a “synthetic but due panel” methodology that makes it it has a to construct to sample attrition, seem like been datasets” the same “synthetic household found that as they panel”the that makesare are called fairly hard it seem like the to come same by in many developingo household relies fundamentally were observed in two time periods. The approach relies fundamentally on time-invariant individual and household characteristics. Specifically, it entails predicting consumption of h rviving households are no longer representative were observed of the in twoastime nation a whole. periods. The approach rounds three relies As of thefundamentally nameAsNLSS period that suggests, the name t-k using onhave the the been“returns” time-invariant proposed suggests, estimated completed the individual methodology proposed to have an and overcomes these time-invariant integrated methodology these pan sho overco characteristi household characteristics. Specifically, it entails predicting consumption of households surveyed in time t for household “returns” characteristics. Specifically, it in entails sampled predicting imputation based methodology to construct a “synthetic panel” that makes it households consumption of “predicted” have been ofconsumption households tracked in t-k over surveyed with to thetime, in time actual but due tafor consumption to insampleperiod taf an imputation based methodology construct “synthetic pane period t-k using the estimated to these time-invariant characteristics period t-k . The comparison s the name suggests, the proposed methodology overcomes these shortcomings and builds on of “predicted” consumption in t-k with period t-k the actual using the consumption estimated period t forms in “returns” to the thesebasis of thewere surviving time-invariantanalysis observed households ofmobility, in are characteristics were two no particularly observed timelonger periods. movement in representative period two The int-k time and . approach The periods. of poverty. out of comparison The relies the fundamentally nation approach as a whole relies on fu mputation based methodology mobility, ANNEX particularly to movement construct 2: inaand “synthetic out of poverty. panel” that makes it seem like the same household household characteristics. Specifically, it entails predicting consumption of ho ere observed in two time periods. of The “predicted” approach relies consumption fundamentally in t-k on with the actual individual time-invariant consumption period and t-k To in period household using write the t more characteristics. formally, estimated forms the basis consider “returns” two of Specifically, to rounds these the of analysis it entails cross time-invariant ofpredicting sectional surveys (deno characteristics con To write more The formally, synthetic mobility, consider two panel particularly rounds of methodology cross movement sectional surveys in and (denoted and out as of application round As the poverty.1 and 2) and name let � to and Nepal suggests, period �� be the t-k the proposed using corresponding the estimated consumption methodology “returns”for the two to overcomes these roundstime-invarian these (these are cons ousehold characteristics. Specifically, it entails predicting consumption of households surveyed inof time t for “predicted” � consumption in t-k with the actual consumption in period t fo imputation i, butbased of methodology wesuppress those subscripts). to For a given a construct “synthetic poverty line, say,panel” z, weconsumptio arethat mak interested “predicted” consumption in and outin of t-k with the actual and �� be the corresponding consumption for the two rounds (these are consumption for household eriod t-k using the estimated “returns” to these time-invariant characteristics in period t-k. The comparison mobility, particularly movement poverty. suppress those subscripts). ToForwritea given poverty more line, say, zconsider formally, , we are interested in estimating were observed fraction ofpoor (a) thesectional in households two (denoted time inperiods. the first The round 1approach who escaped relies poverty (Pr(fundamentall �� > ����� � � “predicted” consumption in t-k with the actual consumption poor households in the first round who escaped poverty (Pr(�� > ����� � ��) or remained in period t two forms rounds the basis of of cross the in analysis poverty in the of surveys mobility, second round, particularly (Pr( � � round as movement ���� � �� ); (b) and in the 2) and and out fraction let of poverty. of � � non-poor household and be the household To write characteristics. more formally, Specifically, consider two � � it entails rounds of predicting cross isectional but consumption wesurveys (denot o obility, particularly second movement round, A (Pr( proper�� and in � study ����out� � of ��� of �poverty. economic ); (b) thecorresponding mobility fraction of non-poor that consumption is households be- thein thefor fractionthe first roundtwoof who rounds non-poor became (thesehouseholds poor To are (Pr( write � consumption � more that ����� > re- formally, �� ) for or the consider household fraction two of ,non-poor rounds of households cross sectiona tha suppress those subscripts). For a given poverty period line, say,(Pr( and t-k �using z,� be the the estimated corresponding� � “returns” consumption �we are interested in estimating (a) the fraction of to these for the time-invariant two rounds (these character are consu poor (Pr(�� ing � > ��) has proposed fairly demanding data require- thatmained non-poor � >������� > ��). �� be.the However, � ����� or the fraction of non-poor households remained non-poor However, this cannot be done in a straightforward manner be of suppress “predicted” �those and subscripts). consumption corresponding inFor t-k given the a with consumption poverty line, say,for z, we the two are inrounds interested o write more formally, ������� > ��). consider However, ments. two Atthis rounds poor the cannot of households minimum, cross be done in sectional it ais in surveys the first to straightforward necessary (denoted round be who manner as because this round escaped cannot 1 of the fact and poverty bethat 2) done and wein let (Pr( doanot � �observe straightforward > ���� the same � manner �� household ) or in the twoactual remained inperiods. poverty consumption in the period �z poor � suppress households in � thethosefirst subscripts). round who For escaped a given poverty poverty (Pr( � line, > say, ���� � , we �� d �� be the correspondingobserve the same consumption household able to observe in for the the two theround, two periods. same household rounds (these in are at leastconsumption because for mobility, household of the fact i , particularly but that we we do movement not observe in the and out of poverty. � ppress those subscripts). two For a given second povertyand line,the (Pr( say,richer � � z, we are � ���� � interested � �� ); (b) in estimating the fraction (a) the in of non-poor second round, fraction Instead poor of periods. households (Pr( we apply �� � ����� in households the in the synthetic � the first panel ��); first(b) the round methodology round fraction who who of inbecame the escaped following non-poor poverty steps: households (Pr(� Instead we apply the time periods synthetic panel methodology in the the following amount steps: same household the two oor households in the first round who poorescaped (Pr( �� � ����� poverty (Pr( �� > > ������) or � the �� ) or fraction remained of To in non-poor write poverty poor more in(Pr( households the �formally, STEP �second � �����1: round, Estimate � that consider > ��the remained )(Pr( or � twothe relationship� rounds non-poor fraction ���� between � of �� cross non-poor ofconsumption ); (b)(Pr( �� sectional the > time households fraction and surveys of that non-p invariant (d of information available on the household, � � the � � STEP 1: Estimate the ������ relationship > �� between ). However, consumption andthis timecannotinvariant be done characteristics in a in straightforward each ������ round > and �� ). obtain poor However, manner estimates (Pr( � this because for � �� cannot ����� and of ��> be the for�� done i )=fact1,or in 2. thata the straightforward we fraction do notof manner non-poor bec h cond round, (Pr(�� � ���� � � �� richer the ); (b) the�fraction analysis that can of be non-poor done. These households “pan- in the Instead first we round and apply ��the who be the became synthetic � correspondingpanel methodol- � consumption � � � for the two rounds (these are co obtain estimates for � �� and � � for i = 1, 2. observe the same household in the two periods. oor (Pr(�� � ������ > ��) el datasets” or the fraction observe as they of � the are same household called non-poor arehouseholds fairly hard in the thattwo to periods. ogy remained in the suppress following non-poor (Pr( those steps: ������� > ��). �� > subscripts). ForHowever, a giventhis poverty cannot �be �� line, � ��done �say, � ��� zin �, we a straightforw are(1)interes ����� > ��). However, this comecannot by be in donemanyin developing a straightforward �� � � �� � countries. � � � manner In the (1) because of the fact that poor householdswe do Instead we apply not observe in the the the first same synthetic round household panel who escaped methodology in the two poverty periods. (Pr(��steps: > ����� � �� �in � �the following � � Instead � �� � � �� (2) bserve the same household case in the of two Nepal, periods. the we three apply rounds the of synthetic NLSS that panel methodology STEP 1: Estimate in thethe following relationship steps: between con- �� � � �� �� � � �� (2) second round, (Pr( Instead � � � the ����apply werelationship � �� � the synthetic ); (b) the fraction panel methodology of non-poor househ the follow ininvariant have been completed have an integrated panel sumption and time STEP 1: Estimate invariant characteristics in between consumption and time ch stead we apply the synthetic panel methodology component, STEP i.e., a subset1: Estimate in the of the the followingsampled relationship steps:house- between consumption each round and obtain poor (Pr( � and obtain estimates �time ����� > invariant � �� ) or characteristics the fraction in each of non-poor round and households estimates for �� and � � for i = 1, 2. � � � STEP STEP 2: 1:Use Estimate the estimates the relationship of � � � and between consumption and holds have obtain been tracked estimates over for time, ��� and but � � � for due toi = 1,i2. = 1, 2. ������� > ��). However, this cannot be done inthe error term a straightforward to predict period manner TEP 1: Estimate the relationship � surveyed in period obtain estimates for �� and � 2. Since � we do not � � for a i� priori= 1,� know2. the empirical distributio sample between attrition,consumption and time the sur- characteristics inobserve thatinvariant each for round the and STEP 2: Use the estimates of �� and the error term to predict period 1 consumption households ��� �� � �� (1) it has been found in period 2. Since we do not a priori know the empirical distribution of the correlation between �� and �� we consider two extreme cases and use the two scenarios to define same household in the two periods.� btain estimates surveyed for � �� and � � viving � for i = 1, 2. households are no longer representative � upper��� � �� � � �� ofmobility. (1) In one case, we assume there and �� we consider two extreme cases and use the two scenarios to define the � and lower bound �� �is �� zero � �� � � correlation �� �� � � �� �� between (2)� �the �� erro mobility. In one of the case, nation we assume as a there whole. is zero correlation between the error terms. In this Instead case, the we incomeapply the synthetic prediction for the first panel round methodology is done by randomly in the following drawing steps: with replacemen nd application to Nepal � � � �� �� � � �� (1) Annex 2: The synthetic �� the � �panel �� �� methodology � �� �distribution and application (2)first-round to Nepal prediction for the first round � is done by randomly drawing with replacement from empirical of the estimated residuals for each household �� � �i� in �the second rou � � � � �� of the first-round estimated As the name � residuals suggests, for each the household proposed i in the methodol- second round. Thus, the STEP period 1 1: Estimate predicted household the relationship consumption forbetween each household consumption surveyed inand period time 2 can invaria be writ � � � �� �� � � �� (2) 2 can be written as: STEP 2: Use the estimates of � � and the error term to predict period 1 household consumption for each household ogy overcomes these shortcomings and builds surveyed in period STEP 2: Use obtain the estimates estimates for of � �� and and� � � for the er- idata = 1, � 2. fairly demanding data requirements. At the A proper study of economic mobility that is being proposed surveyed has in fairly period demanding 2. Since we do not requirements. a priori know At the the empirical distribution on an imputation based methodology to con- ror to term to the same period predict in at least twofor 1 consumption d in at least two time periods and the STEPricher 2:minimum, Use the it is necessary estimates of to be � able and observe the error term and to household � we predictSTEP consider 2: period two Use 1 the extreme time periods and� consumptionestimates cases and of use for the richer households and twothe error term to p struct a “synthetic panel” the that amount makes of it seem information like � � available households on the surveyed household, �in the period richer the 2. Since analysis we that docan be �� � done. �� These �the ��� � � � �� “panel scenarios to (1)define 2: Use that the analysis TEP the can be done. estimates of These � “panel mobility. In one surveyed case, inwe assume period 2. there Since is wezero do correlation not a priori between know the the error empi � � and the error surveyed term in period to predict period 1 consumption for households eloping countries. the same household were datasets” observed as 2. they Since in are two we calledtime do are not fairly a priori not hard toknow a priori come know the by the in empirical many empiricaldevelopingdistribution distribution countries. of ofthe In the correlation case of Nepal, between the with �� rveyed 2:in period 2.In theperiods. Since case we do ofnot Nepal, The a theapplication priori approach and �� we know the relies consider empirical fundamentally two distribution extreme on of the the correlation correlation prediction between between � for and the � first we we round consider consider is done two by � extreme randomly � � � cases � drawing � and � � use thereplacement two(2) scena Annex The synthetic panel methodology and three rounds to Annex Nepalof 2:AnnexNLSS The that cases synthetic have panel and been use theand completed methodology two have of the scenarios � anfirst-round integrated application define to estimated to Nepal � panel component, the upperi.e., residuals for and � a subset each lower � household � ofboundthei in of � the second roun ated panel component, d �� we consider two extreme i.e., a subset casesmobility. time-invariant of the use the andindividual two and scenarios household to 2: define char-The synthetic the upper two paneland extreme methodology lowercases bound and and use of application mobility. the two In to scenarios Nepal one case, to we assume there is zero correlation bet In one case, sampled households we assume have been there is zero tracked over correlation time, household but due between to sample consumption the attrition, error forterms.each it has In been household thisfound case, surveyed thein that income theperiod 2 can be writte ample obility. attrition, In one it has case, we been assume found acteristics. there thatis the correlation zero Specifically, it entails between predicting the error con- terms. define In this the case, upper the and incomelower prediction bound of for the mobility. first round is done by randomly drawing wi f economic mobility that is being proposed prediction has A fairly demanding proper for of surviving study the first households data economic round requirements. mobility are isno done that Atlonger the is randomly byrepresentative being proposed has drawing of the fairly withas nation demanding replacement a whole. data requirements. from At the empirical distribution the a whole. for the first round ediction is sumption done by of randomly households A drawing propersurveyed study with in of replacement time economic t for from mobility Inthatthe one is empirical case, being we proposeddistribution assume has there fairly of the is zero demanding first-round correlation data estimated requirements. residuals At the for each household i in t cessary to be able to observe the same household theinfirst-round of minimum, at least it is two time periods estimated and the residuals richer for each household i in the second round. Thus, � thethe period 1 predicted the first-round ormation available estimated period on the household, residuals the for t-k using richer each the the household estimated As the minimum, analysis necessary that ican “returns” name it inis be to be able the second suggests, necessary done. to these to to These observe round. bethe able “panel Thus, proposed to the same between observe the thetheSTEP household period methodologysame error 1 in at 2: predicted household least terms. Use overcomes two in the In time household at this least periods estimates thesecase, two and the consumption shortcomings time the ofin- periods richer � � and andfortheeach and builds richer error householdon term to predict an surveyed peri in period s these consumption ousehold shortcomings and builds household onthe an amountconsumption of information available for each on the household household, surveyed the richer the surveyed in period analysis in that period can 2analysis can 2.be be done. Since writtenTheseas: we bedo “panel not a priori know the empirical distrib are called are fairly hard to for each household time-invariant come by in many developing surveyedthe characteristics datasets”imputation as in period amount countries. they are called inof based In information period the 2are can case t-k be methodology fairlyof written available The . Nepal, hard tothe to come as: on come the household, construct prediction by in many a the “synthetic developing for richer thepanel”the countries. first round that In the that makes is can case of done it seem Nepal, done. by like These “panel the the same household hat makes it seem like the same comparison householdof “predicted”datasets” wereof as they are consumption observed in called two intimeare t-k thehard fairly periods.randomlytoThe come by in� drawing and many we withdeveloping considerreplacement countries. two from extremeIn the the case cases of Nepal, and usethe the two and scenarios to def NLSS that have been completed have an integrated three panel rounds component, NLSS thati.e., have a subset been of completed have an approach integrated � relies panel fundamentally component, i.e., a on subset time-invariant of the individual amentally lds have been ontracked time-invariant withindividual over time, thedue but actual and sample tosampled three consumptionattrition, household rounds itin has of been periodNLSS beenfound characteristics. that forms ttracked have that been the Specifically, completed empirical it entails have an integrated distribution to predicting of the it consumption panel first-round component, households offoundestimat- i.e., a subset thesurveyed of the time t for between the e in correlation households have over time, but due mobility. sample attrition, In one case, has been we assume thatthere is zero mption lds are no of longer households surveyed representative the ofbasis the in time nation of the t analysis as for a whole. sampledof mobility, households are noparticularly haverepresentative been tracked ed over residuals time, for but due to a each sample attrition, household i in it has the second been found that the surviving period households t-k using the longer estimated “returns” of theto these nation as time-invariant whole. characteristics in is period t-k.randomly The comparison surviving households are no longer representative prediction of the nation for the as a whole. first round done by drawing with replacem haracteristics in period t-k . The movement comparison in and out of of poverty. “predicted” consumption in t-k round. with the Thus, actual the period consumption 1 predicted in period household t forms the basis of the analysis of ggests, the proposed methodology overcomes As the these name shortcomings suggests, the and builds on proposed an methodology overcomesof thethese first-round shortcomings estimated and builds residuals on an for each household i in the second period t forms nmethodology the basis to construct of the analysis a “synthetic panel” of makes that mobility, As the it seem particularly name likesuggests, the same movement the proposed household ain consumption and out panel” of poverty. methodology for each it overcomes household these surveyed shortcomings in builds on an and imputation based methodology to construct “synthetic householdthat makes consumption seem like the for same each household household surveyed in period 2 can be w two time periods. The approach To write reliesmore fundamentally were formally, observed onconsider imputation in two based time-invariant time two methodologyrounds individual periods. Theto and period construct approach a2“synthetic relies can be written fundamentally panel” on as:time-invariant that makes it seem like the and individual same household teristics. Specifically, it entails of cross sectional predicting consumption household To were surveys of write observed (denoted households characteristics. moresurveyed informally, astwo round Specifically, in consider time itperiods. time t for two 1entails The rounds predicting approach of cross consumption relies sectional fundamentally of households surveys on (denoted surveyedtime-invariant in time as round individual t for 1 andand 2) and let �� rveys (denoted as round and he estimated “returns” to these 1 and 2)2)and time-invariantandlet let �� t-k and household characteristics � inbe the characteristics. period corresponding t-k . The Specifically, comparison consumption it� � entails for � the predicting �two � rounds consumption (these of are consumption households surveyed for in household time t for i, but we period and using�the the correspond- beestimated “returns” to these �time-invariant � � = �� �� �characteristics �� (3) in period t-k. The comparison nsumption in t-k with the actual consumption of in period “predicted” t period suppressforms t-k using the those consumption basisthe of subscripts). inestimated the t-k analysis with “returns” For the of a actualgiven � � to � these poverty �� = �� �� � �� consumption � in � time-invariant line, � say, t period z characteristics , we forms (3) arethe interested basis inof period the in t-k analysis . estimating Theof comparison (a) the fraction of se are consumption for household ing consumption i, but we for the two rounds (these are rly movement in and out of poverty. mobility, of “predicted” particularly poor households movementconsumption in and the first round insuppress in out t-k of with poverty. the actual who escaped consumption poverty consumption in period (Pr(�� > ����� from t forms the � ��)equa- basis of the analysis or remained in poverty in the of e interested in estimatingconsumption (a) the fraction forof household i, but we Using the predicted mobility, particularly movement in and out of poverty. > ����� � ��) or remained in poverty Using mally, consider two rounds those of cross subscripts). sectional in Tothe surveys For write a given second (denoted more poverty round, as round formally,consumption (Pr(1line, � and consider two rounds �say, 2) ���� and z , let �� tion �� ); (b)(3),the we can fraction obtain of estimates non-poor of households movements in the let first round who became the predicted � fromof � cross sectional � equation (3), we surveyscan (denoted obtain estimates as round 1 of 2) and andmovements �� in and out of wetwo interested arerounds in estimating poor (Pr( �(a) �the �����fraction > �� of ) or in the and fractionout of of poverty. non-poor For example, households thethat fraction remained non-poor � (Pr( esponding consumption for households in the first round who becamethe (these and Using poverty. � arebe �the To consumption the For write predicted example,more corresponding forformally, household � consumption consumption the fraction� consideri , but from two forwe of poor the rounds two equation of rounds households cross (3), we (thesesectional incan are surveys consumption theobtain first round (denoted estimates for as household round of movements that escaped i , 1 but and povertyinwe 2) and inand let out the �second of �� > bscripts). For a given povertypoor line, say,households z, we suppress in the are interested those ������ and first � in round subscripts). be >estimating �� the ). whoFor However, corresponding(a) aescaped thegivenfraction this poverty of cannot consumption of line, poor be say, done for z thehouseholds , wein two are a interested straightforward rounds in (these the in arefirst estimating manner round consumption (a) the that because fraction for es-of household of the facti , that but we we do not seholds that remained non-poor (Pr( �� > is For poverty. round given example, �� by: the fraction of poor households in the first round that escaped poverty in the second n the first round who escaped poverty poverty (Pr(�� poor > ���� � ��) or households suppress observe in or theremained the remained those first same in in round subscripts). household pov- who poverty Forescaped ina the in given the caped poverty twopoverty poverty (Pr( periods. line,�� say,> in the ���� z � � , we second ��)interested are or round remained inis ingiven poverty estimating by:in (a)the the fraction of manner because of the fact round � is given by: ( �� � ���� � � ��); (b) the fraction ertythat inofthe we second non-poor do not second round, round, households (Pr( in the poor households �� � first���� round in � �� � the who ); (b) first roundthe fraction became who escaped of� non-poor poverty� (Pr(�� > in � households ���� the � first��round who became ) or remained in poverty in the Instead we Pr(� � � �� � �the �) following (4) � steps: � apply the synthetic panel methodology poor (Pr( )� or the fraction � in ���� > ��) or the fraction of thenon-poor fraction householdsof non-poor � � �that second ����� households remained round, > (Pr(��non-poor in � � the ���� first (Pr( � � ���� > );of (b)non-poor the�fraction households � � non-poor households in the first round of that remained non-poor (Pr(�� > who became ������ ). However, this cannot be done in a Pr(� � � �� ��manner straightforward � �) because (4) of the fact that we do not wever, this cannot be done inround a straightforward who became � > manner �� poor poor because(Pr(� of � � the �����fact � that > �� we ) or do not the orhere. fraction of are non-poor households that remained distribution non-poor (Pr( > �� estimated steps: in the two periods. Two things STEP are 1: important Estimate the to note relationship Since between we consumption drawing and from time the empirical invariant characteristics in each of round and household observe the ������same > household �� ). However, in thethis twocannot periods. be done in a straightforward manner because of the fact that we do not Two errors, things we are need � obtain estimates important to repeat to this for �� and� note procedure here. i= Since R number 1, 2. we areof drawing times and from take the the empirical average of distribution equation (4) of to estimated get the observe same household the synthetic in� � � for the two periods. ehe synthetic characteristics invariant panel methodology in in each round the following Instead and measure errors, steps: we we of apply mobility. need theto repeat Second, panel this we methodology assume zero procedure R thecorrelation in number following of steps: times between and take the error the average terms in of(1) and (2) (4) equation theto measure get the of measure STEP mobility of Instead 1: Estimate obtained mobility. we apply by Second, the relationship the this procedure syntheticwe between assume panel is the zero methodology upper �correlation � � � in bound, � the � � � �between following�or � � consumption and time invariant characteristics in each round andmaximum the steps: (1) errormobility. terms We in (1) can and make (2) assumption the measure the relationship between consumption and time invariant characteristics in each round and on of theestimates mobility obtain other extreme obtained for � �� and by – � � that this the for iprocedure = 1, 2. correlations is the upper of the bound, idiosyncratic or maximum shocks are mobility.perfect We and can positive make assumption – and add 75 or � �� and � � for i = 1, 2. � STEP 1: Estimate the relationship between consumption (1) � � � � � � and � � �time invariant (2) characteristics in each round and more on the“persistence” other extreme and – that “stickiness” the correlations to the vector of the of � consumption. idiosyncratic � � � shocks This are would perfect give and uspositive the lower – and bound add obtain estimates for � �� and � �� for�� i� = �� 1, � 2. (1) �� � � �� �� � � �� estimate of mobility.and (1) “persistence” In this case, instead �� of � �� what is of written in (3) we would estimate the period 1 predicted more “stickiness” to �the vector consumption. This would give us the lower bound � = movements ��of � � � (3)out of �� �� � �� in and �� Using the predicted consumption from equation (3), we can obtain estimates poverty. For example, the fraction of poor households in the first round that escaped poverty in the second M O V I N G U P T H E L A D D E R : P O V E RT Y round is given by: R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L Using the predicted consumption from equation (3), we can obtain estimates of movem � For example, poverty. � � the fraction of poor households in the first round that escaped p Pr(� round � � is �� � given � � �) by: (4) Two things are important to note here. Since we are drawing from the empirical Pr(�� � � �� distribution � � � � � �) of estimated (4) errors, we need to repeat this procedure R number of times and take the average of equation (4) to get the measure of mobility. Second, we assume Two zero things are important to note here. Since we are drawing from the empirical di correlation between the error terms in (1) and (2) the measure Two things areofimportant to note here. Since we errors,give would we need us to repeat the lower this procedure bound R number estimate of Wemo-of times and take the average of mobility obtained by this procedure is measurethe upperof bound, mobility. or maximum Second, we assumemobility. zero correlation can make the assumption are drawing from the empirical distribution of on the other extreme – that the correlations bility. In of this the case, instead idiosyncratic of what shocks is are written perfect in between and positive error terms in ( – and add We of mobility obtained by this procedure is the upper bound, or maximum mobility. estimated errors, more we “persistence” need to repeat andthis proce- “stickiness” (3) to thewe on the would vector other of estimate extreme the period consumption. This 1 – that the correlations predicted would of thegive us the lower idiosyncratic shocks bound are perfect a dure R number of times estimate and take In of mobility. the average this consumption case, instead more for each “persistence” of what is written and household in“stickiness” (3) we would surveyed to the vector of estimate in consumption. the This would giv period 1 predicted of equation (4) to get the measure of mobility. period consumption for each household surveyed in period 2 as: 2 estimate as: of mobility. In this case, instead of what is written in (3) we would estimate Second, we assume zero correlation between consumption for each household surveyed in period 2 as: the error terms in (1) and (2) the measure of � ��� =� � � �� � �� � �� (3) ���� =� � � � �� � �� �� (3) mobility obtained by this procedure is the up- per bound, or maximum where � � � mobility. We can make where where � �� is the predicted residuals from (2) � is the In predicted residuals fromnot(2) above. In this case, we would not be draw � is the predicted residuals from (2) above. this case, we would be drawing from an empirical the other extreme assumption ondistribution – that the cor- above. In distribution but case, this we using actual would predictednot be drawing residuals for every household thus we would n but using actual predicted residuals for every household thus we would not have to perform R- relations of the idiosyncratic shocks are perfect replications. from an empirical distribution but using actual replications. and positive – and add more “persistence” and predicted residuals for every household thus we “stickiness” to the vector of consumption. This would not have to perform R-replications. 76 M O V I N G U P T H E L A D D E R : P O V E RT Y R E D U C T I O N A N D S O C I A L M O B I L I T Y I N N E PA L 77 The World Bank Group Nepal Office P.O. Box 798 Yak and Yeti Hotel Complex Durbar Marg, Kathmandu, Nepal Tel.: 4236000 Fax: 4225112 Email: infonepal@worldbank.org www.worldbank.org/np www.facebook.com/WorldBankNepal