Pathways to Reducing Poverty and Sharing Prosperity in India Lessons from the Last Two Decades Urmila Chatterjee, Rinku Murgai, Ambar Narayan and Martin Rama © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Pathways to Reducing Poverty and Sharing Prosperity in India Lessons from the Last Two Decades Urmila Chatterjee, Rinku Murgai, Ambar Narayan and Martin Rama ABSTRACT India is uniquely placed to help reduce global poverty and boost prosperity. The country has the largest number of poor people in the world, as well as the largest number of people who have recently escaped poverty. There is an emerging middle class but the majority of people are still vulnerable to falling back into poverty. What lessons do the past two decades offer for what it will take for the country to sustain progress and bring about deeper changes? This synthesis brings together the key insights from extensive and in-depth research conducted by the World Bank on India’s experience in reducing poverty and sharing prosperity. The first chapter offers an overview of the trends in living standards and mobility in India. This is followed by a chapter on the main drivers of poverty reduction. The third chapter sheds light on some of the gaps India needs to fill for sustaining mobility and spreading prosperity more widely. Acknowledgements: Carlos Felipe Balcazar Salazar, Hai-Anh Dang, Basab Dasgupta, Gaurav Datt, Sonalde Desai, Hanan Jacoby, Peter Lanjouw, Yue Li, Gaurav Nayyar, Monica Yanez Pagans, Swati Puri, Martin Ravallion and Christina Wieser contributed to the research underlying this paper. We thank the Indian Express for partnering with us in disseminating this research to its readers through a series titled “Tackling poverty in India”. Comments and guidance by Benu Bidani, Ana Revenga and Onno Ruhl, from the peer reviewers Abhijit Sen and Luis-Felipe Lopez Calva, and participants at various seminars and workshops are gratefully acknowledged. The authors may be contacted at uchatterjee@worldbank.org. | | i i     P A T H WAY S TO R ED U CI N G PO VER TY A N D SH A RING PROSPE RIT Y IN INDIA    Contents 1. Trends in Poverty 1 Poverty has declined at an increasingly rapid pace 1 Prosperity could have been shared more widely 2 There was substantial upward mobility but a majority remains vulnerable 4 Progress on non-monetary dimensions of wellbeing was uneven 7 Some population groups fared substantially worse 9 India’s Poverty Profile 11 2. Drivers of Poverty Reduction 14 Poverty is increasingly concentrated in low-income states 14 No particular sector of activity was more pro-poor in its growth 15 Cities, more than specific sectors, drove poverty reduction 17 Jobs, more than transfers, mattered for households 18 Tackling Poverty in India: The Indian Express series 21 3. Sustaining Mobility and Sharing Prosperity 22 Not enough (good) jobs are being created 22 Demographic dividend versus declining female labor force participation 24 A paucity of good locations 26 Locations in the mid-range of the rural-urban gradation do converge 28 The economic forces behind rapid convergence can be enhanced 30 References33 Data Annex 34 | Contents    i i i  rends in 1. T Poverty Poverty has declined at an increasingly rapid pace India has made tremendous progress in reducing lifted out of poverty. Moreover, the pace of absolute poverty in the past two decades. The poverty reduction accelerated over time and was standard way to determine whether a household three times faster between 2005 and 2012 than is poor is to compare its daily expenditure per in the previous decade. Poverty rates fell at a capita to a minimum consumption threshold, or similar pace in rural and urban areas, although a poverty line. Based on India’s official line, the vast majority of the poor (four out of every five) share of the population living in poverty was still live in rural areas. halved between 1994 and 2012, falling from 45 percent to 22 percent (figure 1). During this International metrics validate this positive story. period, an astonishing 133 million people were Based on a globally comparable poverty line set Figure 1: Poverty has declined rapidly, especially in recent years Annual change in poverty rate (%) 0 1994 to 2005 2005 to 2010 2010 to 2012 -1 -2 Note: Based on National Sample -3 Surveys (NSS). Consumption is expressed in constant 2005 All India Rural Rupees, corrected -4 for cost-of-living differences between states and rural and urban areas using India’s -5 official poverty lines. Rural Urban Total Source: Narayan and Murgai (2016). 1. T rends in Poverty    1 | The pace of poverty reduction is now faster than elsewhere Figure 2:  Population below poverty line (%) 50 46.1 40 34.7 30 20 21.3 10 14.1 0 1993 1996 1999 2002 2005 2008 2011 India Lower middle income Middle income Developing World Note: Based on the international poverty line of $1.90 per day (in 2011 Purchasing Power Parity). Figures are available at roughly 3-year intervals during 1990-2008. Data are from the NSS for India, and from World Development Indicators (WDI) for other countries. Source: Narayan and Murgai (2016). at $1.90 per person per day (in 2011 Purchasing (figure 2). As a result, India’s share of the global Power Parity), India accounts for the largest extreme poor declined from 30 percent in 2005 to number of people that have escaped poverty in 26 percent in 2012. However, despite the enormous recent years. After a lackluster performance in progress poverty remains widespread. One in every the 1990s, the pace of poverty reduction in India five Indians is poor, nearly 270 million people. exceeded that of the developing world as well as And, at the global poverty line, India is home to that of Middle Income Countries (MICs) as a group the largest number of poor in the world today. Prosperity could have been shared more widely The inclusiveness of economic growth can be times faster towards the end of the period than it assessed based on the growth rate of per capita had been at the beginning. But despite the four- consumption among the bottom 40 percent of the fold increase, it still lagged behind the growth in population. This indicator of shared prosperity consumption for the population as a whole. improved significantly after 2005, tracking the poverty trend closely (figure 3). The growth in India’s rather unremarkable performance in consumption for the bottom 40 percent was four sharing prosperity with the bottom 40 percent | | 2     P A T H WA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    Consumption has grown faster on average than at the bottom Figure 3:  Annual growth in consumption per capita (%) 6 5.6 5.2 5 4 3 2.1 2 1.7 1.3 0.8 1 Note: Consumption 0 expressed in constant 1994 to 2005 2005 to 2010 2010 to 2012 2005 All India Rural Rupees. poorest 40% Average Source: Narayan and Murgai (2016). of its population contrasts sharply with its solid among the richest segments of the population performance in terms of average consumption based on household surveys. The latter do a growth (figure 4). Between 2005 and 2012 good job at capturing relatively basic forms of India ranked 16th among 51 MICs based on consumption, but are not well-suited to quantify the consumption growth rate of the overall fanciful expenditures such as trips abroad or population, but it only ranked 27th based on luxurious housing. Moreover, the rich are less the consumption growth rate of the bottom likely to spend time responding to surveys of 40 percent of its population. this kind than the poor, which leads to under- reporting at the top of the distribution. These This assessment is not inconsistent with a are possible reasons why India’s average growth relatively stable degree of overall inequality. in household consumption as measured by A standard indicator in this respect is the Gini household surveys lags systematically behind index, which varies from 0 in a situation of perfect the growth of private consumption as measured equality to 100 percent in the hypothetical through national accounts. situation in which one household accounts for the entire income or consumption of the An alternative way to assess the inclusiveness country. During this period India’s Gini index has of economic growth is the elasticity of remained stable at around 32 percent, which is poverty reduction to economic growth, or the relatively low by international standards. But the percentage change of the former when the Gini index considers the entire population, and latter increases by one percentage point. In can remain stable if inequality among the bottom this indicator, poverty is measured based 40 percent or the top 60 percent declines while on household surveys but economic growth inequality between the two groups increases. is measured based on national accounts, implicitly correcting for the under-measurement This said, the assessment is tainted by the of household expenditures among the difficulty to adequately measure consumption non-poor. | 1. T rends in Poverty    3 India’s economic growth was not especially inclusive Figure 4:  12% Annual growth in per capita consumption/income of bottom 40% 10% Brazil China 8% (2005-2010) (2007-2012) Vietnam 6% (2004-2010) Thailand (2008-2012) Russian Fed 4% (2007-2012) South Africa Turkey (2006-2011) (2007-2012) 2% Sri Lanka India (2006-2012) (2005-2012) 0% Nigeria Note: For Mexico, Brazil, -2% (2003-2009) Germany and Italy, income growth figures are used; -4% consumption growth figures are used for all other named countries. Data are from the -6% Global Database for Shared -6% -4% -2% 0% 2% 4% 6% 8% 10% 12% Prosperity, at the World Bank. Annual growth of per capita consumption/income of total population Source: Narayan and Murgai (2016). This other measure confirms that India’s growth Encouragingly, growth seems to be becoming has not been particularly inclusive in recent years. more inclusive over time. The elasticity of For the period from 2005 to 2012, its elasticity of poverty reduction to economic growth more poverty reduction to economic growth ranks in than tripled from 1994-2005 to 2005-2012, the 35th percentile among the 116 developing with much of the improvement occurring in countries for which data are available. Put the last two years of this period. In 1994- differently, in roughly two thirds of developing 2005, one percentage point of economic countries growth was more inclusive than in India growth brought about a 0.24 percent reduction during this period. This relatively low elasticity in the poverty rate at the $1.90 line. By is the reason why despite India being among the 2005-2012, the corresponding decline in the top performers in terms of economic growth it poverty rate had accelerated to 0.93 percent. was just above the 60th percentile of developing And it had reached an impressive 2.24 percent countries in the rate of poverty reduction. in 2010-2012. There was substantial upward mobility but a majority remains vulnerable The rapid reduction in poverty means that there poverty line than there were households falling were many more households moving above the below it. But the dynamics were similar at various | 4     P A T H WA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | levels of expenditure per capita, and not just among those households that were poorer in around the poverty line. And movements upward 2005 (figure 5). Households that were better- were more frequent among the poor than among off in 2005 experienced slower consumption the non-poor. growth, with some taking the place of the poorest households by 2012. This “churning” of To assess the extent of mobility it is necessary households moving up and down relative to other to go beyond aggregates such as the poor, or households explains why the anonymous growth the bottom 40 percent, and track the trajectory rates for poorer households are much lower than of individual households. In other words, it is the non-anonymous ones. necessary to shift from “anonymous” to “non- anonymous” measures of wellbeing. The NSS, Mobility can also be assessed through transitions which is the source of consumption expenditure of households over time between well-defined data used for producing official poverty estimates population groups – such as the poor, the in India, does not allow for this, except through vulnerable and the middle-class. This other statistical approximations, but the India Human approach also points to high upward mobility. Development Survey (IHDS) does. Based on the Its implementation required to first define in a IHDS, between 2005 and 2012, the consumption rigorous manner the dividing line between the of an average Indian household grew at about vulnerable and the middle class. In practice this 4.7 percent per year. An anonymous measure was done by choosing a threshold for expenditures suggests that the growth rate was roughly the per capita such that households above it would same at every percentile of the distribution. face a probability of falling into poverty lower But a non-anonymous measure, which compares than 20 percent. Based on this metric, more than consumption per capita of the same households half the population changed group from 2005 to between the initial and final years, shows 2012, and more than two thirds of those changing that consumption growth was much faster group moved upward (figure 6). Poorer households were more likely to move up Figure 5:  Growth incidence curves, consumption 20 Annual growth in consumption per-capita -10 0 10 0 20 40 60 80 100 Note: Based on IHDS. Consumption per-capita percentiles Consumption and incomes are expressed in All-India Rural Non-anonymous anonymous 2005 Rupees. Source: Balcazar et. al (2016). 1. T rends in Poverty    5 | Figure 6: There was high mobility, with upward movements dominating Middle - class 6.7 14.6 1.5 2005 Vulnerable 8.2 18.2 13.8 Poor 15.3 15.9 5.7 0 10 20 30 40 50 Note: Based on a synthetic % of total population panel constructed out of two NSS rounds. 2012 Poor 2012 Vulnerable 2012 Middle-class Source: Dang and Lanjouw (2015). Figure 7: A middle-class is rising, but a persistently large vulnerable group remains Share in total population (%) 50 40 41 40 37 34 30 25 23 20 10 0 Note: Based on a synthetic 2005 2012 panel constructed out of two NSS rounds. Poor Vulnerable Middle -class Source: Dang and Lanjouw (2015). Strong upward mobility was enough for the the vulnerable group and not into the middle- Indian middle-class to grow into the second class. As a result, the vulnerable continued to largest segment of the population by 2012 – a be the largest population group (around 40 full third of it – as befits India’s emergence as percent of the population) over the period. Many a middle-income country during the last decade households that escaped poverty after 2005 still (figure 7). However, most of those who escaped had consumption levels that were precariously poverty between 2005 and 2012 moved into close to the poverty line in 2012. | 6     P A T H WA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Progress on non-monetary dimensions of wellbeing was uneven The poverty status of a household is assessed were higher in Nepal, Bangladesh and Cambodia based on its daily expenditure per capita under than in India, but they were lower in 2014. the assumption that the household can buy the goods and services it needs. But for some basic A particular area of concern remains services there may not be a market. Households undernourishment among children. Some Indian may lack access to electricity, or to sanitation, or states, including a few high-income ones, show to health services. A comprehensive assessment stunting and underweight rates that compare of the progress made in raising living standards poorly with the averages for low-middle income needs to take into account these non-monetary countries, sub-Saharan Africa, and some of the dimensions of wellbeing as well. other countries in South Asia. While there are multiple forces at play, the prevalence of diarrheal Consistent with the reduction in monetary disease is thought to be one of the main reasons poverty, non-monetary indicators of welfare have behind these high levels of malnutrition, and also improved steadily in India over the last two diarrhea is triggered by poor hygiene. In 2015, decades. But they have done so to a lesser extent 60 percent of the Indian population lacked than in other developing countries. In some access to improved sanitation, and 44 percent cases, countries that had human development practiced open defecation. Both shares are indicators at comparable levels in the early- higher than in Bangladesh, Nepal and Pakistan, 1990s are doing better by now (figure 8). For despite all three countries having lower income instance, in 1994, child and infant mortality rates levels. Figure 8: Infant mortality declined more slowly than in comparable countries Under - 5 mortality rate 120 100 India, 112 India, 75 80 60 India, 50 40 20 1994 2005 2014 Note: All figures are in terms of per 1000 live Nepal Bangladesh Cambodia births. India Vietnam Nicaragua Source: Narayan and Murgai (2016). 1. T rends in Poverty    7 | The extent of non-monetary deprivations Such uneven progress in the different dimensions varies not only across countries, but also of wellbeing needs to be taken into account within countries. There is a strong correlation when assessing the “true” speed of poverty between household consumption per capita reduction. For instance, in India the share of and access to basic services, reflecting the fact households with access to electricity is similar that richer households can afford to move to across small and large rural areas, or across small better neighborhoods, or may have more clout and large urban areas, but urban areas as a whole to bring public services to the places where have substantially higher access. Household they live. But there is also a strong correlation expenditures, on the other hand, grow quite between access to services and urbanization. Not steadily across the four types of locations, from surprisingly, urban households tend to have both less to more urban places (figure 9). Therefore, higher consumption levels and better access to the same increase in household expenditures services than rural households. But monetary is associated with a stronger improvement in and non-monetary dimensions of wellbeing do wellbeing when it results from moving from rural not necessarily improve at the same rate as rural to urban areas than when it arises from moving areas urbanize. up within each of the two groups. Figure 9: Access to electricity was strongly associated with urbanization 98% 2500 100 93% 2221 79% 76% Access to electricity (% households) 2000 80 Expenditure per capita (Rupees) 1599 1500 1409 60 1229 1000 40 500 20 Note: Small rural comprises 0 0 villages with less than 5,000 Small rural Large rural Small urban Large urban inhabitants; large urban comprises cities with more than one million inhabitants. Real per capita expenditure Access to electricity Source: Authors, based on NSS 2012. | 8     P A T H WA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Some population groups fared substantially worse Living standards among specific population groups from Scheduled Castes and especially from have consistently lagged behind the rest of the Scheduled Tribes. A greater share of Scheduled country. Households belonging to the Scheduled Tribes than other groups have stayed poor in Tribes and Scheduled Castes stand out for not just 2005 and 2012, indicating higher levels of entrenched poverty, but also more deprivation on chronic poverty (figure 11). non-monetary dimensions of wellbeing such as health and education. These groups are sizeable: Differences in non-monetary dimensions in 2012, Scheduled Tribes accounted for 9 percent of wellbeing between these disadvantaged of India’s population and Scheduled Castes for groups and the rest of the population are 19 percent. At 43 percent, Scheduled Tribes have considerable as well. Fewer adults from the highest poverty rate among all social groups, Scheduled Tribes and Scheduled Castes have twice as high as the India average (figure 10). completed secondary school; nearly two in Moreover, poverty has declined at a slower pace every five are illiterate (figure 12). In addition, among Scheduled Tribes. these two disadvantaged groups have lower access to drinking water in their homes and While upward mobility was widespread after practice higher rates of open defecation than 2005, it was more limited among households other groups. Figure 10: Poverty was higher, and declined more slowly, among Scheduled Tribes Population below poverty line (%) 70 60 Pace of 60 poverty reduction 51 50 43 -5% per year 38 40 - per year 29 8% 30 23 21 -8% per year 20 12 -8% per year 10 0 2005 2012 Scheduled Tribes Scheduled Castes Other Backward Castes General Source: Authors, based on NSS. | 1. T rends in Poverty    9 Figure 11: Scheduled Tribes enjoyed less upward mobility and were more vulnerable Social group by transition category, 2005-2012 (%) Scheduled Tribes 32 8 32 28 Scheduled Castes 16 9 31 44 Other Backward Castes 10 7 28 55 Others 4 5 15 76 0 20 40 60 80 100 Stayed poor Became poor Became non-poor Stayed non-poor Source: Authors, based on IHDS. Figure 12: Disadvantaged groups fared worse on non-monetary dimensions of wellbeing Education attainment, 2012 (% adults) Access to basic services, 2012 (% households) 100 100 80 80 60 60 40 40 20 20 0 0 ST SC OBC Others ST SC OBC Others Illiterate Literate or primary Open Defecation Middle school completed No drinking water on premises Secondary school or higher completed Note: ST stands for Scheduled Tribes, SC for Scheduled Castes, and OBC for Other Backward Castes. Source: Authors, based on NSS. | 1 0     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | India’s Poverty Profile SNAPSHOT 2012 270,000,000 = 1 in 5 Indians is poor Indians are poor THE LOW-INCOME 60 THE 7 LOW-INCOME STATES HOUSE STATES ARE HOME TO 62 % OF INDIA’S PO OR 45% OF INDIA’S POPULATION 24 UTTAR PRADESH 36 MADHYA PRADESH Number of BIHAR poor in low-income 80 % of India’s poor 10 13 live in rural areas RA JASTHAN states JHARKHAND (Millions) 10 14 CHHATTISGARH ODISHA Poverty Rate Poverty Rate 25 % 14 % in rural areas in urban areas 27 poor % Small Villages pop: 0-4999 19 poor % 17 % Big Villages pop: 5000+ poor Small Towns 6 % poor pop: 0-1mn POOR NON-POOR Big Cities pop: 1mn+ 1 May 20, 2016 Poverty is highest among scheduled tribes SCHEDULED SCHEDULED OTHER BACKWARD OTHERS OTHERS ST Only 28% of Indians are SC CASTES CASTES TRIBES and ST (ST) (SC) (OBC) SC OBC ST OTHERS SC But 43% of the poor are SC and ST OBC 43 % 29 % 21 % 12 % poor poor poor poor Casual labor is the main source of Self employment and casual labor is income for the rural poor the main source of income for the urban poor CASUAL LABOR NON-FARM 17 % 12 % CASUAL LABOR NON-FARM 34 % 10 % CASUAL LABOR FARM 34 % 18 % SELF-EMPLOYED NON FARM 12 % 17 % SELF EMPLOYED NON-FARM 40 % 34 % SELF-EMPLOYED FARM 30 % 36 % SALARIED 4% 10 % SALARIED 20 % 44 % OTHERS 5% 6% OTHERS 7% 12 % The poor spend more on food, fuel and light POOR NON-POOR EDUCATION FUEL & LIGHT EDUCATION FUEL & LIGHT & HEALTH 13 % & HEALTH 9% 6% 11% 25% POOR 33 % NON - POOR 47 % OTHERS 56 % OTHERS FO OD FO OD 2 May 20, 2016 The poor own fewer assets POOR NON-POOR 61% 86% 29% 65% 27% 61% 5% 29% 2% 24% 0% 11% 0% 7% 0% 5% MOBILE TV STOVE TWO REFRIGERATOR WASHING PC / LAPTOP MOTOR CAR PHONE WHEELER MACHINE / JEEP Secondary school completion In rural areas, more marginal POOR is low among the poor land owners among the poor NON-POOR LANDLESS 5% 6% 15 % SECONDARY & ABOVE 37 % 15 % MARGINAL ( <1 HA ) 82 % 72 % MIDDLE 25 % 17 % LITERATE OR PRIMARY 20 % SMALL ( 1-2 HA ) 9% 11 % 45 % ILLITERATE 26 % MEDIUM & LARGE ( >2 HA ) 5% 10 % POOR NON-POOR The poor have lower access to basic services LATRINES ELECTRICITY TAP WATER POOR 21 % 61 % 6% NON-POOR 62 % 85 % 33 % 3 May 20, 2016  rivers of Poverty 2. D Reduction Poverty is increasingly concentrated in low-income states Poverty is not only more prevalent among specific so-called low-income states are Bihar, population groups, such as the Scheduled Tribes: Chhattisgarh, Jharkhand, Madhya Pradesh, Orissa, it is also highly concentrated in specific locations. Rajasthan and Uttar Pradesh (figure 13). As Seven of the 36 states and union territories a result, low-income states as a group – with account for 45 percent of India’s population Rajasthan as the exception – have a poverty rate but nearly 62 percent of its poor. These that is twice that of the rest of the country. Figure 13: A growing share of India’s poor live in low-income states Bubble size: number of poor (millions) 25 UP 20 share of poor > share of population State share in India's poor, 2012 (%) BH 15 MP 10 MH share of poor < share of population OD WB JH 5 KA CG RJ AS GJ AP TN Note: Nineteen large states HR are considered. Low-income KL 0 HPUK PJ states are highlighted in 0 5 10 15 20 25 orange. State share in India's population, 2012 (%) Source: Authors, based on NSS and Population Census. | 1 4     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Moreover, these low-income states are yet to Poverty reduction in the low-income states catch up with the rest of the country in growth has also not been as responsive to economic and poverty reduction. Between 2005 and 2012, growth as in the other states. Admittedly, with the exception of Bihar and Rajasthan, the these states did experience greater absolute low-income states grew at a slower pace than reductions in poverty in the period from 2005 the rest of the country (figure 14). This lack of to 2012. However, measuring catch-up using convergence is a salient characteristic of India, absolute changes can be misleading, given relative to other major federal entities. The US and that initial levels of poverty and per capita the European Union operated as “convergence incomes differed vastly across states. In machines”, gradually bringing poorer members of relative terms, there has been divergence in the federation closer to the living standards of both growth and poverty reduction across richer ones. Indian states. Figure 14: Low-Income States are not only poorer: they also grew more slowly 12 UK 11 10 Strengthening Leading Annual growth rate, 2005-2012 (%) 9 TN 8 AP GJ MH KL HR 7 HP BH RJ AI OD KA 6 MP CG PJ 5 UP WB JH 4 Lagging Weakening AS 3 Note: Nineteen large Indian states are 2 considered here. Low-income states are 0 5000 10000 15000 20000 25000 30000 35000 40000 45000 highlighted in red. Real GSDP per capita, (2005 Rupees) All India Source: Authors, based on data from Central Statistical Office (CSO). No particular sector of activity was more pro-poor in its growth Knowing that poverty reduction was faster them more successful. An obvious candidate is outside low-income states is not enough to the composition of their economic growth by understand what about those other states makes sector of activity. Indeed, the sharp decline in 2. Drivers of Poverty Reduction    1 5| poverty observed in India in recent years, and the construction sector alone accounted for nearly considerable upward mobility associated with it, half of the expansion in non-farm employment occurred against the backdrop of rapid structural (figure 15). In a somewhat surprising way, this transformation. construction boom was felt more in rural areas and especially among the unskilled. With most India’s economic growth is increasingly driven new jobs being created outside of agriculture, by the secondary and the tertiary sectors. in 2012, for the first time more than half of the Between 2005 and 2012 the share of total people at work in India were not on the farm. output contributed by agriculture declined from 19 percent to 14 percent. The contribution of Structural transformation also took the form services increased from 53 to 57 percent, whereas of greater integration, reflected in stronger the share of manufacturing remained relatively inter-sectoral linkages. Growth in one sector stable. now transmits its gains elsewhere to a greater extent than in the pre-liberalization era (before Structural transformation was quite dramatic 1991). Back then rural growth, especially in the when assessed from an employment point of view. farm sector, was what mattered most for poverty Nearly 34 million jobs in agriculture were lost reduction. But in recent times, it is more difficult between 2005 and 2012. In parallel, employment to attribute poverty reduction to the performance in the non-farm sector grew at an annual rate of of any specific sector. The impact of an additional 3.6 percent, adding about 50 million jobs. The percentage point of growth on the poverty rate is Figure 15: Farm employment declined rapidly while most new jobs were in construction Number of jobs (mn) 2005 2012 Annual job growth, 2005-2012 (%) FARM Farm (FARM) -2 MANU Manufacturing (MANU) 2 THR Trade, hotels and restaurants (THR) 2 CONS Construction (CONS) 10 PUB Public and community services (PUB) 2 TRAN Transportation (TRAN) 4 FIRB Finance, real estate and business (FIRB) 6 MINE+UTIL Mining and utilities (MINE+UTIL) 4 0 50 100 150 200 250 300 Source: Authors, based on NSS and Population Census. | | 1 6     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    the same, regardless of the sectoral composition thirds of the post-1991 poverty reduction, and of that growth. From that perspective, poverty the secondary sector about a quarter. But this decline has become sector-neutral. is simply because the non-farm sector accounts for a larger share of GDP and grows In absolute numbers, the contribution of the faster than the farm sector. It is not due to non-farm sectors towards poverty reduction is growth in the non-farm sector being intrinsically by now larger than that of the farm sector. The more pro-poor than growth in the rest of the tertiary sector alone has contributed nearly two- economy. Cities, more than specific sectors, drove poverty reduction In parallel with structural transformation, the Total population, population density and the pace of urbanization picked up. Urban population share of employment in non-farm activities are increased by 32 percent between 2001 and 2011, the three criteria used by the Census of India to almost double the percent increase in total classify a locality as urban. But many localities population. For the first time ever the absolute which are considered urban based on these increase in population was larger in urban areas. indicators are still rural from an administrative This rapid urbanization process has been messy point of view. The rapid multiplication of these in nature. Part of it is the result of urban sprawl, hybrid “census towns” shows that the boundaries with rural areas densifying and gradually being between rural and urban areas have become subsumed into nearby cities. blurred (figure 16). By now, there is no longer Figure 16: Urban population growth is faster in administratively rural areas Number of towns Population 5000 4041 3894 2001 2011 Increase 4000 3799 2987 3000 Total 1030 mn 1210 mn 18% 2000 1702 1362 All Urban 286 mn 377 mn 32% 1000 Statutory Towns 265 mn 323 mn 22% 0 Census 1991 Census 2001 Census 2011 Census Towns 21 mn 54 mn 157% Statutory Towns Census Towns Source: Authors, based on Population Census. | 2. Drivers of Poverty Reduction    1 7 a rural-urban divide in India, but rather a rural- poverty reduction as a whole. This reflected urban gradation. the weak linkages between cities and the rural economy. Post- 1991, rural growth, though still The growth of cities, which encompasses both important, has been displaced by urban growth bigger population and higher productivity, has as the most important contributor to even faster been good for overall poverty reduction in India. poverty reduction (figure 17). Put differently, the In the pre-1991 period, while urban growth poor living in rural areas have gained more from reduced urban poverty, it contributed little to urban growth than from rural growth. Figure 17: Urban growth contributed more to poverty reduction in recent years Pre-1991 Post-1991 .2 .2 Change in log headcount index (with controls) Change in log headcount index (with controls) .1 .1 0 0 -1 -1 -2 -2 -3 -3 -.02 -.01 .01 -.03 -.02 -.01 0 .01 .02 -.03 0 .02 Share-weighted change in log urban mean Share-weighted change in log urban mean Note: Based on NSS. Source: Based on Datt et al. (2016). Jobs, more than transfers, mattered for households The effects of the economy-wide structural shift towards non-farm activities. This shift was transformation manifested at the household level more noticeable among households that escaped in the form of more non-farm jobs and higher real poverty. Jobs in the non-farm sector were mainly wages. As a result, there was a diversification of created by the construction sector. These jobs income sources, especially for households living were far from ideal in terms of regularity in in rural areas. While agriculture continued to wage payments, job security, or social protection be important for many, there were fewer days coverage. But they offered higher earnings spent working on the farm and a significant compared to farm labor. | 1 8     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Labor earnings, from both self-employment The reason why labor earnings played such an and wage employment, on average accounted important role was the unprecedented rise in for nearly 90 percent of household income in real wages for unskilled labor between 2005 and 2012. But in addition, changes in labor earnings 2012 (figure 19). The dynamism of construction were a more significant contributor to higher activity, together with higher minimum support expenditures per capita than other changes prices and favorable terms of trade in agriculture, simultaneously affecting households (figure 18). resulted in higher labor demand both in the These other changes concern remittances and farm and the non-farm sectors. The expansion transfers – as when a household gains access to a in schooling, together with a decline in rural social protection program – and the composition female labor force participation, slowed down the of the household – for instance when a young growth in labor supply. These two forces led to a family member marries and moves out. These tightening of the market for unskilled labor and a other factors did contribute to raising living steep rise in the wages of casual workers. standards. The share of transfers and remittances in household incomes increased considerably As a result, the rural-urban wage gap has narrowed between 2005 and 2012, even if it remained small considerably, especially at the lower end of the overall. And the share of household members distribution (figure 20). This wage compression who work increased, as could be predicted in a contributes to blurring the distinction between country undergoing a demographic transition. rural and urban areas and reinforces the But the change in labor earnings remained by far hypothesis of a growing rural-urban integration the main contributor to poverty reduction. of the Indian economy. Figure 18: Non-farm wage employment was the main ticket out of poverty By sector By type of job All Rural Urban All Rural Urban 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Non agricultural activities Wage/salaried work Other sources of income Other sources of income Composition of households Composition of households Agricultural activities Self-employed work Residual Residual Note: Sources of poverty reduction. Based on IHDS, 2005 and 2012. Source: Balcazar et al. (2016). | 2. Drivers of Poverty Reduction    1 9 Figure 19: Rural wages increased dramatically during the last decade Annual growth in real wages of rural men (%, 2005 Rupees) 50 40 30 20 10 0 1998-1999 1999-2000 2000-2001 2001-2002 2002-2003 2003-2004 2004-2005 2005-2006 2006-2007 2007-2008 2008-2009 2009-2010 2010-2011 2011-2012 2012-2013 2013-2014 2014-2015 -10 Sowing Picking Mason Construction Workers Source: Authors, based on Reserve Bank of India (RBI). Figure 20: Urban-rural wage gaps are closing, especially at the bottom Urban-rural gap in real wages .1 .2 .3 .4 .5 .6 .7 .8 .9 1 1.1 1.2 0 0 10 20 30 40 50 60 70 80 90 100 Percentile 2005 2012 Source: Authors, based on NSS. | 2 0     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Tackling Poverty in India The Indian Express series Five key drivers of reducing poverty in India May 17, 2016 – Onno Ruhl and Ana Revenga India completes 25 years of the beginning of economic reforms this July. Starting today, The Indian Express will publish findings from an e-symposium that brings together recent research by The World Bank on poverty in India. On poverty and prosperity, lot done, lot to do May 18, 2016 – Ambar Narayan and Rinku Murgai The rapid decline in India’s poverty levels over the last decade augurs well for the country’s efforts to eradicate poverty. Poverty down, but 1 in 2 hangs by a thread May 25, 2016 – Peter Lanjouw and Rinku Murgai Scheduled Tribes stand out as a group that has fallen further behind, with one-third stuck in chronic poverty. 1 in 3 has piped water, 2 of 5 kids stunted May 27, 2016 – Ambar Narayan and Swati Puri Decline in consumption poverty notwithstanding, on parameters like tackling open defecation, India lags behind even Bangladesh, Nepal and Pakistan. The low income, low growth trap June 7, 2016 – Urmila Chatterjee and Swati Puri Bihar, Chhattisgarh, Jharkhand, Madhya Pradesh, Odisha, Rajasthan and Uttar Pradesh continue to lag behind the rest of the country in income and growth. Despite the success of these states on a few important fronts, where you live still determines how well you live. India, the driver of growth for Bharat June 13, 2016 – Gaurav Datt, Martin Ravallion, and Rinku Murgai Since 1991, 80% of the total reduction in poverty has been due to urban growth — rural poor have gained more from urban growth than from rural growth. Also, post-1991, secondary and tertiary sectors have helped more to reduce poverty than primary sector. Jobs, not transfers, the big poverty-buster June 17, 2016 – Carlos Felipe Balcazar, Sonalde Desai, Rinku Murgai and Ambar Narayan Between 2005 and 2012, structural changes drove poverty reduction — non-agricultural incomes rose the fastest, and the largest shifts from farm to salaried non-farm employment were seen among the poorest. Where you live decides how ‘well’ you live June 28, 2016 – Yue Li and Martin Rama ‘Good’ living spots tend to be found in clusters; some ‘good’ locations spread more prosperity than others. Three jobs deficits in unfolding India story July 25, 2016 – Martin Rama, Urmila Chatterjee and Rinku Murgai The quantity and quality of jobs created raise concerns about the sustainability of poverty reduction, and the prospects for enlarging the middle class. Since 2005, fewer jobs for women in India July 26, 2016 – Urmila Chatterjee, Martin Rama and Rinku Murgai After farming jobs collapsed post 2005, alternative jobs considered suitable for women failed to replace them, leading to women withdrawing from the labor force. Key lessons on road to sharing prosperity August 17, 2016 – Martin Rama A review of India’s experience over the last two decades confirms the links between poverty and the lack of assets at the household level, but the importance of location suggests interventions by policymakers to go beyond simply investing in education and health.  ustaining Mobility 3. S and Sharing Prosperity Not enough (good) jobs are being created The rapid decline in poverty during a time of high to turn the much awaited demographic dividend economic growth between 2005 and 2012 was into a demographic curse. fueled to a large extent by an expansion in non- farm employment, mainly in the construction A second important deficit concerns the sector, combined with an unprecedented increase quality of the jobs that were created during in real wages for unskilled labor. Strong growth this period. Employment growth took place may well be sustained over time, but some of mainly in construction, where jobs tend to be the factors that contributed to the increase casual. Their wages are set on a daily basis or in real wages may not. The global super-cycle through short-term contracts, and there is no in commodity prices seems to have halted, and job security or social protection associated domestic prices for agricultural products has with them. As a result, the shift of employment already caught up with international prices, out of agriculture has been associated with meaning that there is little scope to see farm- an increasing casualization of non-farm work. gate prices increasing much. Casual jobs help people escape poverty in the short run, but they do not guarantee entry into While labor earnings grew rapidly, the number the middle class. This sectoral composition of of jobs did not. In fact, the period 2005-2012 changes in employment is, thus, consistent with can be described as being characterized by a the high levels of vulnerability of households growing jobs deficit. Or rather three of them. to falling into poverty observed between 2005 The first one concerns the absolute numbers. and 2012. Between 2005 and 2012, net job growth in the economy was 0.6 percent per year. This was Transitions into the middle class are associated much less than the growth in the working age with wage employment. The likelihood of a population that was not in school – 1.9 percent household durably escaping poverty between per year. In absolute numbers, out of the 2005 and 2012 was higher if a larger share of 13 million potential entrants into the workforce its members had regular jobs (figure 21). On the every year during this period only 3 million got other hand, the share of family members holding a job. In a young and increasingly aspirational casual jobs increased among households that society, this growing jobs deficit has the potential slipped into poverty between these two years. | | 2 2     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    Figure 21: Regular jobs support a more durable escape from poverty Types of jobs for those who were non-poor in both 2005 and 2012 Types of jobs for those who where non-poor in 2005 but poor in 2012 Non-farm 28% Non-farm 12% regular 28% regular 11% Non-farm 12% Non-farm 18% casual 16% casual 26% Non-farm self 15% Non-farm self 11% employed 15% employed 8% 32% 39% Farmers Farmers 30% 35% Farm casual 10% Farm casual 18% 8% 17% 2005 2012 2005 2012 Source: Authors, based on IHDS. Figure 22: Regular jobs are predominant only in large urban areas Types of jobs, 2012 (% employed) 100 7 14 80 38 34 55 40 60 18 40 8 58 46 44 20 37 0 Small rural Large rural Small urban Large urban Source: Authors, based on Self-employed Casual wage Regular salaried NSS and Population Census 2001. In principle, urbanization brings with it the substantially higher in large urban areas. But promise of better jobs. And in the case of there are much fewer regular jobs in small towns, India, it is true that the share of regular jobs is and they are rare in rural areas (figure 22). 3. Sustaining Mobility and Sharing Prosperity     2 3 | Higher wages for the unskilled in rural areas and of regular jobs in large villages and small towns, a massive transition out of farming supported where most of the Indian population lives, the rapid poverty reduction observed in recent building a large middle class will remain an years. But in the absence of a vibrant creation elusive goal. Demographic dividend versus declining female labor force participation The third jobs deficit characterizing the period Labor Force Participation Rate (LFPR) exceeded 10 2005-2012 was the shortage of suitable jobs for percentage points during this period (figure 23). women. One of the most striking developments The decline was particularly pronounced in rural during this period was the decline in the share of areas, where the female LFPR fell from 49 percent of working-age women who work or actively seek work. the working-age population in 2005 to 36 percent Precise numbers vary depending on the definition in 2012. The rate remains relatively stable in urban of employment used, as some activities performed areas, but at a very low level as only one in five by women – especially at home, on a non-regular working-age women living in cities is economically basis – could be treated as self-employment, active. As a result of this downward trend, India inactivity or unemployment. But regardless of today is near the bottom in female LFPR among the definition used, the decline of the female countries with similar income levels. Figure 23: Female labor force participation has declined sharply in rural areas Female labor force participation in rural areas, 2012 (%, age 15+) Female labor force participation in urban areas, 2012 (%, age 15+) 50 50 40 40 30 30 20 20 10 10 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Principal Status Usual Status Principal Status Usual Status Current Weekly Current Daily Current Weekly Current Daily Note: Based on NSS. Source: Chatterjee et al. (2015 b). | 2 4     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    | Poverty fell rapidly in India between 2005 and earnings in real terms, as was roughly observed 2012, but it would have fallen even faster had between 2005 and 2012, would lead to a decline female LFPR remained constant at its 2005 level. in female LFPR by about 3 percentage points. Since then, many rural households lost out on the This rough estimate is corroborated by a much earnings of their female members who became more careful analysis matching characteristics inactive. Beyond short-term living standards, of women’s households with those of the places economic inactivity undermines agency by they live in. women, and slows down progress towards gender equality. Gainful work by women, and especially A more plausible explanation has to do with the paid employment, is correlated with their agency increasing scarcity of “suitable” jobs for women. at the household level and in society more In a traditional society, women’s work is more broadly, and with better development outcomes, acceptable if it takes place in environments including greater investments in children’s health perceived as safe and provides enough flexibility and education. The male LFPR, on the other hand, to simultaneously perform household duties has remained high at about 80 percent in both and chores. Working in the family farm matches rural and urban areas. this description, and indeed female LFPR is high in small villages, where agriculture remains the A common explanation for the decline in female main economic activity. Work outside the family LFPR is the expansion in access to secondary house is also more acceptable if it takes place in education. Girls are staying longer in school, hence a relatively protected environment, such as an working less at younger ages. This is a welcome office or a factory. But in recent years the number development, both from a skills perspective and of farm jobs has dropped dramatically in India, from a gender equality perspective. However, this without a parallel emergence of regular jobs in explanation can only account for a fraction of the offices and factories. observed decline. Most of the observed decline in female LFPR actually occurred among older In rural areas, the only non-farm jobs available women. And it took place in spite of their higher in large numbers are in construction, and they educational attainment. Among women aged involve casual work. Men employed in this sector 18 to 30 years, the share of those completing worked mainly for private contractors or on their secondary education increased from 20 percent own account. By contrast, more than half of the in 2005 to 32 percent in 2012. But for the same women working in construction in rural areas age cohort, the share in the labor force declined were doing so under MGNREGA and other public from 38 to 30 percent. works programs. MGNREGA alone accounted for over a third of the female construction workers in A second explanation focuses on the so- rural areas in 2012. called “income effect”. It is argued that in a predominantly patriarchal society the relative The scarcity of suitable jobs for women has prosperity of recent years has allowed more become particularly marked in the rapidly- women to stay at home, a preferred choice for expanding areas that are neither truly rural their husbands. This explanation is plausible, nor fully urban. Between 2005 and 2012, farm but on closer examination it can only account for jobs collapsed in the villages, whereas regular about a fourth of the decline in female LFPR. It is employment only expanded significantly in large true that female LFPR fell more in districts where urban areas. The combination of these two trends labor earnings increased more substantially. But created a “valley” of suitable jobs for women the relationship is such that a doubling of labor along the rural-urban gradation (figure 24). | 3. Sustaining Mobility and Sharing Prosperity     2 5 Figure 24: There are not enough suitable jobs for women along the rural-urban gradation Types of jobs, 2012 (% female adults) 20 Valley of Suitable Jobs 15 10 5 0 Small rural Large rural Small urban Large urban Farmers Non-Farm Self Non-Farm Regular All Casual Note: Based on NSS and Population Census 2001. Source: Chatterjee et al. (2015 b). A paucity of good locations The jobs deficits experienced by India during correlates of poverty. Traditionally, attention the period 2005-2012 are strongly linked with has focused on household endowments and the urbanization process. Regular employment other characteristics as the most important grew mainly in large urban areas, whereas the determinants of poverty. For instance, shortage of “suitable” jobs for women was felt households with lower educational attainment more strongly in large rural areas. Scheduled tend to be poorer. But even controlling for a Tribes, the group that was more clearly left large range of household characteristics, nearly behind during this period, are also concentrated a third of the variation in living standards across in specific districts, and live mainly in small households can be attributed to their place of rural areas. These observations call for a deeper residence. understanding of the spatial patterns of mobility and exclusion. A greater spatial granularity is Building on this insight, it is possible to compute especially pertinent in the case of India, where the location premium associated with more than states are massive entities. 1,400 places along the rural-urban gradation in more than 600 Indian districts. This location When defined at a fairly disaggregated level, premium (positive or negative) is measured location appears as one of the most important as the additional expenditure per capita an | | 2 6     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    average household would enjoy, relative to the it lives. Places with high location premiums tend average place in India. The focus is on nominal to be close to each other, forming clusters of high expenditure, in current Rupees, which means living standards. These clusters are most often that the premium may partially reflect higher situated around a top urban location, but they can prices, and not fully translate into higher living spread out over a vast catchment area with still standards. However, higher nominal expenditure substantially high location premiums. Catchment is usually associated with higher earnings, and areas encompass both urban and rural places. earnings increase with labor productivity. This Many of these clusters and their catchment areas makes the location premium a defensible measure include high-performing villages. of productivity in a particular place. The best places do not share their prosperity Not surprisingly, urban places perform better than evenly, however. For instance, both Bangalore rural places and large urban areas display the and Delhi are among India’s top places. The highest location premiums. But a careful spatial location premium is slightly higher in Bangalore, analysis shows that some of the best places in which suggests that it is a more productive city. India are small towns. The analysis also reveals But households in the catchment area of Delhi do a large degree of overlap in location premiums, substantially better than those in the catchment along the rural-urban gradation (figure 25). At area of Bangalore. The location premium is still the turn of the century, a similar analysis revealed positive and large up to 200 km away from core a much sharper divide between rural and urban Delhi, while it almost vanishes 100 km from core areas. Bangalore. It is not only where a household lives that Places with the lowest location premiums tend to matters for living standards, but also next to what be contiguous as well. They are concentrated in Figure 25: Large villages and small towns have similar location premiums 2.5 Large rural and small urban The best performers are are almost indistinguishable in ‘’small urban’’ 2 1.5 Desnsity 1 .5 0 -1 -.75 -.5 -.25 .0 .25 .5 .75 1 Location permiums Small rural Large rural Small urban Large urban Note: Based on NSS 2012 and Population Census 2001. Norminal consumption based, OLS Source: Li and Rama (2015). | 3. Sustaining Mobility and Sharing Prosperity     2 7 central India and happen to be in many of the a large share of the Scheduled Tribes (figure 26). low-income states. They are mainly rural – but This suggests that social exclusion is closely include few small towns – and they are home to intertwined with spatial exclusion in India. Figure 26: Where one lives, and near what, matters for poverty Top places (70) Catchment places (49) Average places (318) Bottom places (162) No data (34) State District Note: Based on NSS 2012 and Population Census, 2001 and 2011. Source: Li and Rama (2015). Locations in the mid-range of the rural-urban gradation do converge A spatially disaggregated analysis reveals more states generally performing worse than the rest. If convergence in living standards across India than household expenditures per capita are considered, the comparison across states suggested. When instead of GDP, there is neither divergence nor considering states there is divergence in the convergence. A tentative explanation for the growth rates of GDP per capita, with low-income difference between divergence in GDP per capita | | 2 8     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    and neither convergence nor divergence in effort to understand why strong convergence in household expenditures per capita has to do with living standards across places does not translate internal migration. If migrants from low-income into convergence across states. states work in more vibrant states and send remittances to their families, they generate GDP The explanation, again, is related to the where they migrated to, but support consumption urbanization process. Rapid convergence is back home. This said, even when considering happening in the mid-range of the rural-urban household consumption per capita, there is no gradation. Household expenditures per capita evidence that low-income states are catching up. grow faster in large rural and small urban places than in either small rural or large urban places On the other hand, there is absolute convergence (figure 27). There is also convergence within in living standards when the district, rather than each of the four groups, and convergence the state, is the unit of analysis. And the speed is faster among large urban places. All this of convergence is twice as fast when considering suggests that the economic forces that sustain an even higher level of spatial disaggregation, shared prosperity are stronger in more urbanized distinguishing between small rural, large rural, settings, whereas there is divergence at the lower small urban and large urban places. This finding end of the rural-urban gradation. Low-income is not a statistical artifact, driven by higher states may thus be failing to converge because measurement error when considering smaller they have not been as successful at urbanizing places. But the finding warrants some additional as other states. Figure 27: The mid-range of the rural-urban gradation is catching up Annual growth rate of expenditure per capita (percent) 4.00 Large villages grow faster than Large villages and (poorer) small small towns grow 3.75 villages faster than (richer) large cities 3.50 3.25 But there is convergence within 3.00 each of the four groups of places 866 938 1085 1570 2.75 Expenditure per capita in 2005 (Rupees per month, log scale) Small rural Large rural Small urban Large urban Note: Based on NSS 2005 and 2012, and Population Census 2001. Source: Li and Rama (2016). | 3. Sustaining Mobility and Sharing Prosperity     2 9 The economic forces behind rapid convergence can be enhanced Given that the growth in living standards differs such conceptual “buckets”, each with multiple considerably across locations, it is important to indicators. understand what makes some locations perform better than others. Such understanding provides The million regressions approach leads to clues on the kind of policies and investments discarding about half of the indicators that which have the potential to accelerate poverty economic theory, or previous analyses, would reduction and foster shared prosperity. But there have picked up as top candidates to drive growth are two significant methodological challenges in at the local level. The results suggest that the trying to identify the key characteristics of well- most important predictor of subsequent growth performing places. is belonging to an urban cluster, and preferably to one with a large population. Major urban centers The first challenge has to do with internal with vast catchment areas, such as Delhi, share migration. A sending place may be growing their prosperity deep into surrounding places more slowly than a place receiving migrants which can be administratively rural. The second because its population has a shrinking share of most important set of indicators is related to people with characteristics (in terms of age or infrastructure, and includes access to electricity education) that make them more productive, and and density of roads (density of railways, less so). not because the place is becoming less productive Market access, the average distance to places in any fundamental way. To get around this issue, with high levels of economic activity, comes next one would consider convergence in location (figure 28). premiums (rather than convergence in household expenditures per capita) as they refer to an The economic structure of the place also appears average household with the same characteristics to be an important predictor of subsequent in all places across India. economic growth. Places with a larger share of medium-size and large firms grow faster, as do The second methodological challenge has to do places with a more diversified economic structure. with the multiplicity of characteristics that could The share of the local labor force having a regular potentially have an impact on local performance. job also appears to be a strong predictor of rapid Following the literature on convergence, this is growth. Other indicators related to the economic addressed by the “million regressions” approach, structure, such as the share of the construction to assess which characteristics are consistently and manufacturing sectors in total employment, significant correlates of growth in premiums matter as well. But their impact is not as large as at the local level. Economic theory, as well as that of larger firms and regular employment. previous analyses, point to a multiplicity of factors that could make a difference. Governance, Last but not least, inclusion seems to contribute infrastructure, market access, economic to faster local growth. Starting with financial structure, types of jobs, inclusion, human inclusion: places that grow faster had initially a capital and climate are among the potentially larger share of households with access to finance. relevant characteristics to consider. The spatial The same holds true, although to a lesser extent, data available for India allow considering nine for places with a larger share of firms borrowing | | 3 0     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    Figure 28: There are predictors of rapid growth at the local level Growth impact of an increase by one standard Bucket Indicator deviation (percentage points) -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Belongs to a cluster (yes = 1) Cluster population (million) Governance Is a state capital (if yes, percent of state’s population) Is a municipality (if yes, percent of population in municipalities) Road density (kms per sq. km) Households with electricity (percent of households) Infrastructure Firm with electricity (percent of firms) Railway station density (per sq. km) Nightlight-based (weighted distance) Market access GDP-based (weigthed distance) Medium-size firms (percent of firms) Large firms (percent of firms) Economic Diversification (inverse of Herfindahl index) structure Construction (percent of working-age population) Manufacturing (percent of working-age population) Regular wage (percent of working-age population) Types of jobs Casual wage (percent of working-age population) Self-employment (percent of working-age population) Households with bank accounts (percent of households) Firms with formal borrowing (percent of firms) Scheduled Castes (percent of population) Inclusion Gender gap in tertiary education (percentage points) Gender gap in secondary education (percentage points) Gender gap in literacy rate (percentage points) Scheduled Tribes (percent of population) Literacy rate (percent of working-age population) Human capital Primary education (percent of age group) Secondary education (percent of age group) Temperature variability (standard deviation) Climate Precipitation (mms. per year) Note: Based on data from NSS 2005 and 2012 and Population Census 2001. Two statistical criteria are used to decide when to retain an indicator. Darker bars are for indicators meeting the two criteria, lighter bars for indicators meeting only one of them. Source: Li and Rama (2016). 3. Sustaining Mobility and Sharing Prosperity     3 1 | from formal financial institutions. Importantly, educational attainment, grow more slowly. Places various forms of social exclusion appear to be where a larger share of the population belongs detrimental to subsequent growth. For example, to Scheduled Tribes, the population group most places with low literacy rates and primary ostensibly left behind in recent years, also school enrollment, or with large gender gaps in experience slower growth. | | 3 2     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    References Balcazar Salazar, Carlos Felipe, Sonalde Datt, Gaurav, Martin Ravallion, and Rinku Murgai Desai, Rinku Murgai and Ambar Narayan (2016) “Growth, Urbanization and Poverty (2016) “Why did Poverty Decline in India? A Reduction in India.” Policy Research Working Nonparametric Decomposition Exercise.” Policy Paper 7568, World Bank, Washington DC. Research Working Paper 7602, World Bank, Washington DC. Jacoby, Hanan and Basab Dasgupta (2015) “Changing Wage Structure in India in the Post- Chatterjee, Urmila, Rinku Murgai and Martin Rama Reform Era: 1993-2011.” Policy Research Working (2015 a) “Employment Outcomes along the Rural- Paper 7426, World Bank, Washington DC. Urban Gradation.” Economic and Political Weekly Vol 50(26&27). Li, Yue and Martin Rama (2015) “Households or Locations? Cities, Catchment Area and Prosperity Chatterjee, Urmila, Rinku Murgai and Martin in India.” Policy Research Working Paper 7473, Rama (2015 b) “Job Opportunities along the World Bank, Washington DC. Rural-Urban Gradation and Female Labor Force Participation in India.” Policy Research Working Li, Yue and Martin Rama (2016) “The Drivers Paper 7412, World Bank, Washington DC. of Strong Convergence at the Place Level in India”. Unpublished manuscript, World Bank, Dang, Hai-anh and Peter Lanjouw (2015) Washington DC. “Poverty Dynamics in India between 2004 and 2012: Insights from Longitudinal Narayan, Ambar and Rinku Murgai (2016) Analysis using Synthetic Panel Data.” Policy “Looking Back on Two Decades of Poverty and Research Working Paper 7270, World Bank, Well-Being in India”, Policy Research Working Washington DC. Paper 7626, World Bank, Washington DC. | References    3 3 Data Annex Name Description National Sample The NSS is an annual nationally representative household survey covering Survey (NSS) different topics over time. Data on consumption expenditure are collected nearly every year, with larger samples of households interviewed in the “thick” rounds that are used for estimating poverty rates. Detailed information on employment is also collected in the thick rounds. The most recent survey years on consumption and employment are from 2004-05, 2009-10 and 2011-12. The NSS is also the source for non-monetary indicators of well-being such as educational attainment, access to electricity, and open defecation. The NSS is produced by the National Sample Survey Organization. India Human The IHDS is a nationally representative, multi-topic survey of households Development conducted in 2004-05 and 2011-12. The survey has a panel structure, meaning Survey (IHDS) that it is possible to track the same households over time. In addition to gathering information on consumption expenditures and income, the IHDS also covers topics such as employment, education, health, and access to social programs. The IHDS is produced by the National Council of Applied Economic Research (NCAER) and the University of Maryland. Wage and Price These indices are computed on a monthly basis at locations throughout the Indices country. Wage indices provide information on wage growth for men in rural areas. Consumer price indices for agricultural laborers are used to compute real wage growth. Data are gathered by the Central Statistical Organization, and curated and released by the Reserve Bank of India. Population Census Data from the Census of India is used to compute population at all administrative levels. It is also used to determine the class and size of cities, and to estimate urbanization rates. This source allows the identification of sampling substratum in the NSS, serving as the basis for analyses below the district level. The custodian for this data is the Registrar General and Census Commissioner of India. National Accounts This is the source of data on Gross Domestic Product (GDP) at the national levels, as well as for states and districts. Data on GDP per capita is used to measure economic growth, to identify low-income states, and to estimate the elasticity of poverty reduction to economic growth. Growth rates of GDP per capita at the state level are used to assess economic convergence over time. National Accounts are produced by the Central Statistical Organization. | | 3 4     P A T HWA Y S TO R ED U CI N G PO VER TY AN D SH ARING PROSPE RIT Y IN INDIA    Name Description South Asia Spatial This database is a repository of geo-referenced indicators for all countries of Database South Asia. The India module is currently available. It builds on around 30 databases and provides information on a range of socio-economic indicators, including the urban extent, demographics, jobs, economic activity, infrastructure, ICT, finance, business, living standards, education, health and environment. Data are drawn from population and economic censuses, household and firm-level surveys, administrative records, satellite imagery and crowdsourced data. The number of indicators varies with the level of spatial disaggregation. The South Asia Spatial Database is maintained by the World Bank. Global Database on This database includes the most recent figures on annualized consumption or Shared Prosperity income growth of the bottom 40 percent of the population, as well as related indicators. Data are for 94 countries over the period 2007-2012. The figures are estimated based on nationally representative household surveys, and data is curated so that the indicators are comparable across countries. The database is maintained by the World Bank. World Development WDI is a collection of development indicators at the country level, across a range Indicators (WDI) of topics and sectors. Data are compiled from officially-recognized international sources. Indicators are also aggregated at regional and global levels. The WDI database is a product of the World Bank. | Data Annex    3 5