Global Poverty Monitoring Technical Note 2 September 2018 PovcalNet Update What’s New Shaohua Chen, Dean M. Jolliffe, Christoph Lakner, Kihoon Lee, Daniel Gerszon Mahler, Rose Mungai, Minh C. Nguyen, Espen Beer Prydz, Prem Sangraula, Dhiraj Sharma, Judy Yang and Qinghua Zhao September 2018 (Updated March 2019*) Keywords: What’s new; September 2018; India; LIS; API Development Data Group Development Research Group Poverty and Equity Global Practice Group Abstract The September 2018 update to PovcalNet involves several changes to the data underlying the global poverty estimates. Some welfare aggregates have been changed for improved harmonization, and some of the CPI, national accounts, and population input data have been revised. This document explains these changes in detail and the reasoning behind them. Emphasis is given to the updates to the Indian poverty estimates. In addition to the changes listed here, 24 new country-years have been added, bringing the total number of surveys to 1601. All authors are with the World Bank. Corresponding authors: Christoph Lakner (clakner@worldbank.org) and Minh C. Nguyen (mnguyen3@worldbank.org). The authors are thankful for comments and guidance received from Aziz Atamanov, Joao Pedro Azevedo, R. Andres Castaneda Aguilar, Paul A. Corral Rodas, Reno Dewina, Carolina Diaz-Bonilla, Francisco Ferreira, Jose Montes, Laura Moreno Herrera, Rinku Murgai, David Newhouse, and Ayago Esmubancha Wambile, and to Ruoxuan Wu for excellent research assistance. This note has been cleared by Francisco Ferreira. *March 2019 update: • Appendix with the CPI source for each country-year has been added (Table A.1). This refers to the CPIs used for the September 2018 PovcalNet update. The Global Poverty Monitoring Technical Note Series publishes short papers that document methodological aspects of the World Bank’s global poverty estimates. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Global Poverty Monitoring Technical Notes are available at http://iresearch.worldbank.org/PovcalNet/. Contents 1. Introduction ............................................................................................................................. 1 2. Changes to welfare aggregates ................................................................................................ 3 2.1. Bhutan 2003 .................................................................................................................... 3 2.2. EU-SILC data.................................................................................................................. 3 2.3. LIS data ........................................................................................................................... 3 2.4. Fiji 2013.24 ..................................................................................................................... 4 2.5. Kenya .............................................................................................................................. 4 2.6. Malaysia 2008.25 ............................................................................................................ 4 2.7. Rwanda 2010.83 and 2013.75 ......................................................................................... 4 3. Changes to CPI data ................................................................................................................ 5 4. Changes to national accounts data ........................................................................................... 6 5. Changes to population and survey coverage data .................................................................... 7 6. Revisions to India’s poverty estimates .................................................................................... 8 6.1. CPI revisions ................................................................................................................... 8 6.2. Revisions to line-up after 2011.5 .................................................................................... 8 7. Country-years added .............................................................................................................. 12 8. Estimating shared prosperity in China................................................................................... 12 9. Other changes ........................................................................................................................ 13 References ..................................................................................................................................... 14 Appendix ....................................................................................................................................... 15 1. Introduction The September 2018 global poverty update from the World Bank presents new poverty estimates for 2015, and revises the previously published global and regional estimates from 1981 to 2013. The update includes new surveys that have been received and processed, as well as several changes to the existing data. Some changes reflect improvements in the welfare aggregate based on new harmonization efforts and more available information. This document outlines the changes made to the underlying data by country, and explains the reasons why the changes have been made. Table 1 shows the global and regional poverty estimates for 2015, which are presented in more detail in the 2018 Poverty and Shared Prosperity report (World Bank, 2018). In 2015, an estimated 736 million people were living below the international poverty line, currently set at $1.90 in 2011 purchasing power parity (PPP) U.S. dollars. The global poverty rate, the share of the world’s population living below the international poverty line, stands at 10%, while 26% live on less than $3.20 and 46% live on less than $5.50. Sub-Saharan Africa accounts for more than half of the world’s population below the international poverty line and has the highest regional poverty rate, at 41%. Table 1. Poverty estimates for reference year 2015, different poverty lines $1.90 $3.20 $5.50 Survey Head- Num- Head- Num- Head- Num- Region coverage count ber of count ber of count ber of (%) ratio poor ratio poor ratio poor (%) (mil) (%) (mil) (%) (mil) East Asia and Pacific 97.6 2.3 47 12.5 254 34.9 710 Europe and Central Asia 89.9 1.5 7 5.4 26 14.0 68 Latin America and the Caribbean 89.8 4.1 26 10.8 68 26.4 165 Middle East and North Africa 64.6 5.0 19 16.3 61 42.5 158 South Asia 21.4 n/a n/a n/a n/a n/a n/a Sub-Saharan Africa 52.7 41.1 413 66.3 667 84.5 849 Other High-Income Economies 71.7 0.7 7 0.9 10 1.5 16 World Total 66.7 10.0 736 26.3 1933 46.0 3386 Source: PovcalNet Note: Survey coverage is assessed within a two-year window either side of 2015, i.e. including surveys that were conducted between 2013 and 2017 (see section 5 below). The estimates for South Asia are not displayed since the region has a survey coverage less than 40%. 1 Table 2 illustrates the impact of the data updates on global poverty for the reference year 2013. The estimates for 2013 were first published in October 2016, and have since been revised in October 2017 and April 2018. With the new data, the estimate of the global $1.90 headcount ratio increased from 10.9% to 11.2% and the number of poor increased from 783 million to 804 million people. The additional 21 million poor people at the global level can be largely explained by a revision of the line-up methodology in India (see section 6.2 for details), which increases the estimated number of poor in India by 17 million (from 210 to 227), and increases the headcount ratio in South Asia from 15.1% to 16.2%. The remaining change is mostly explained by a new survey in Kenya in 2015.67.1 This new survey adds more than 2 million poor people in Kenya compared to the previous estimate, which was based on an extrapolation of the 2005.38 survey. Apart from India and Kenya, no country had its estimate change by more than half a million poor people. Table 2. Poverty at reference year 2013: Comparison of April and September 2018 versions $1.90: $1.90: $3.20: $3.20: $5.50: $5.50: Headcount Number of Headcount Number of Headcount Number of ratio (%) poor (mil) ratio (%) poor (mil) ratio (%) poor (mil) Region Apr Sep Apr Sep Apr Sep Apr Sep Apr Sep Apr Sep 2018 2018 2018 2018 2018 2018 2018 2018 2018 2018 2018 2018 East Asia and 3.6 3.6 73 73 17.6 17.5 352 352 42.5 42.4 853 852 Pacific Europe and Central 1.6 1.6 7.7 8 5.8 5.8 28 28 14.1 14.1 68 68 Asia Latin America and 4.5 4.6 28 28 11.3 11.4 69 70 27.1 27.2 166 167 the Caribbean Middle East and 2.7 2.6 9.6 9 14.5 14.4 52 51 42.7 42.3 153 152 North Africa South Asia 15.1 16.2 257 274 52.6 53.9 894 916 83.5 84.2 1418 1431 Sub-Saharan 42.3 42.5 401 405 67.5 67.8 639 645 85.2 85.4 807 813 Africa Other High-Income 0.6 0.6 6.4 6 0.9 0.8 9.5 8.9 1.5 1.5 16 16 Economies World Total 10.9 11.2 783 804 28.6 28.8 2044 2072 48.7 48.7 3481 3498 Source: PovcalNet Note: The increase in the number of poor at the $5.50 line, without any change in the headcount ratio, can be explained by rounding and an upward revision of the population total due to the inclusion of Eritrea and Taiwan, China (see Section 5). 1 The decimal year notation is used when data are collected over two calendar years. The number before the decimal point refers to the first year of data collection, while the numbers after the decimal point show the proportion of data collected in the second year. For example, the 2015.67 Kenya survey was conducted in 2015 and 2016 with two thirds of the data collected in 2016. Also see footnote 3 in Atamanov et al. (2018) and Lakner et al. (2018) for details. 2 2. Changes to welfare aggregates 2.1. Bhutan 2003 The 2003 data have been updated with a new version of the data and a revised consumer price index (CPI). With the new welfare aggregate, the Gini index declined from 46.78 to 40.9. Introducing the new aggregate and the new CPI caused poverty at $1.90 to decline from 36.22% to 21.33%. 2.2. EU-SILC data All historical EU-SILC data have been updated to data released in March 2018. The updates for each country-year are documented on the Eurostat website [CIRCABC → Eurostat → EU-SILC → Library → data_dissemination → udb_user_database]. The following country-years were revised (referring to the reference year of the welfare aggregate): Croatia (2015), Iceland (2014), Netherlands (2015) and Sweden (2007-2013). The effects on the poverty estimates and other distributional statistics are minor. 2.3. LIS data The Luxembourg Income Study (LIS) is a database of harmonized microdata from 50 countries from around the world. PovcalNet uses the disposable income variable from the LIS database for seven countries: Australia, Canada, Germany, Israel, Japan, South Korea, and United States. Disposable income is given as the sum of labor income, capital income, public transfers, private transfers, less taxes and contributions. Pending further research on harmonizing the treatment of negative incomes across our database, we exclude households with negative disposable income. Disposable income is expressed in per capita terms without applying equivalence scales (as is the case with all other surveys used in PovcalNet). LIS does not distribute the microdata, so PovcalNet includes grouped data generated from the LIS microdata (via the LIS remote execution system). Since the April 2018 update, the method to generate grouped data for the seven LIS countries has been updated such that it is consistent with how grouped data are generated for the countries that rely on SILC data. In particular, 400 bins are now created instead of 300 bins (with one exception, see footnote), and the bins are created using Stata’s _ebin command, developed by Joao Pedro Wagner De Azevedo, rather than Stata’s 3 xtile command.2 _ebin is similar to xtile, but they differ in how observations with the same income are treated and _ebin generates bins that are more similar in population size. The _ebin command can be downloaded by typing ssc install alorenz. 2.4. Fiji 2013.24 New consumption items were added to the welfare aggregate, reducing poverty slightly. As a result of these changes, Fiji’s poverty rate at $1.90 for 2013.24 changes marginally, from 1.39% to 1.37%. The Gini index for the same survey changes from 36.37 to 36.70. 2.5. Kenya The methodology used to estimate international poverty in Kenya was revised for consistency across rural and urban households. The Kenya National Bureau of Statistics (KNBS) excludes rent expenditure for all rural households. As a result, for national poverty estimation, two poverty lines are used that account for these differences in the aggregate. Since the international poverty line does not allow for differentiation between urban and rural households, rent expenditures are now excluded also for urban households. The CPIs in Kenya have also been revised (see section 3). In 2005, the headcount ratio at $1.90 was revised upwards from 42.8% to 43.6% while the Gini declined from 48.5 to 46.5. In 2015, poverty at $1.90 changed from 35.8% to 36.8%, with the Gini declining from 42.6 to 40.8. 2.6. Malaysia 2008.25 The previously included 2009 data have been replaced with data for 2008.25, reflecting updated information about the reference period of the welfare aggregate. Furthermore, the welfare aggregate has been updated to net income from the Household Income Survey. The Gini index and poverty at $3.20 changed from 46.3 to 45.5 and 3.1% to 4.2%, respectively. 2.7. Rwanda 2010.83 and 2013.75 The 2010.83 and 2013.75 welfare aggregates are now spatially deflated. This has generated moderate adjustments. The $1.90 poverty headcount ratio for the 2010.83 survey increased from 2 When the bin size does not reach a minimum number of observations, the number of bins is iteratively reduced by 50. For this reason, we use 250 bins in Israel 1979. This has minor implications for poverty estimates. 4 60.4% to 62.4%, and the Gini index fell from 51.3 to 47.2. For the 2013.75 survey, the $1.90 poverty headcount ratio fell from 59.5% to 56.0%, and the Gini index fell from 50.4 to 45.1. Since the aggregates before 2010.83 are not spatially deflated, there is now a break in the series. More details can be found in Fatima and Yoshida (2018). 3. Changes to CPI data The baseline source of CPI data has not been updated from the April 2018 vintage of PovcalNet. It remains the IMF’s International Financial Statistics (IFS) as of December 2017. 3 Yet, some changes have been made, primarily to older survey years, where IFS data are not available. Table 3 summarizes the changes to the CPI data as part of September 2018 PovcalNet update. Most of the changes concern using the World Economic Outlook’s annual CPI series as the main secondary source whenever IFS data are unavailable, and monthly CPI data are not needed. Lakner et al. (2018) provide an overview of the various CPI series that are used in PovcalNet in more detail. For countries not listed in Table 3, there were no changes in the CPI data source between the April 2018 and September 2018 versions. Table A.1 in the Appendix to this note gives the current source of the deflator for all countries included in PovcalNet. 3 A few recent surveys require CPIs for 2017, which are not available in the December 2017 vintage of the IFS. This concerns Bhutan 2017, Gabon 2017, Indonesia 2017, Uganda 2016.5, and West Bank and Gaza 2016.85. In these cases, CPIs for 2017 from more recent IFS vintages are combined with the CPI series from the December 2017 vintage. 5 Table 3. CPI data sources: Comparison of April 2018 and October 2018 versions Economy Years Description of change in CPI data Argentina 1986, 1987 Switched to CPI from World Economic Outlook Bangladesh 1983.5, 1985.5, 1988.5 Switched to CPI from World Economic Outlook Bosnia and Herzegovina 2001, 2004 Switched to CPI from World Economic Outlook Belize All years Switched to CPI from World Economic Outlook Bhutan 2003 Corrected an error in IFS data Chile All years until 2006 Switched to CPI from the ILO Micronesia, Fed. Sts. 2000, 2005, 2013 Updated CPI from World Economic Outlook Guinea 1991, 1994.08, 2002.25 Switched to CPI from World Economic Outlook Guyana 1992.5 Switched to CPI from World Economic Outlook India 1983, 1987.5, 1993.5, Updated CPI from the National Statistical Office 2004.5 (see 6.1 for additional details) Iran, Islamic Rep. 1986, 1990, 1994, 1998 Updated CPI from the National Statistical Office Kenya 1992, 1994, 1997 Updated CPI from the National Statistical Office Lao PDR 1992.2 Switched to CPI from World Economic Outlook Lesotho 1986.54, 1994.45 Switched to CPI from World Economic Outlook Malaysia 2008.25 Change in survey year from 2009 to 2008.25, and weighted CPI changed accordingly.1 Mozambique 1996.27, 2002.5 Switched to CPI from World Economic Outlook Namibia 1993.79 Switched to CPI from World Economic Outlook Romania 1989 Switched to CPI from Milanovic (1998) Sierra Leone 1989.75, 2003.25 Switched to CPI from World Economic Outlook Tajikistan 1999 Switched to CPI from World Economic Outlook Timor-Leste 2001 Switched to CPI from World Economic Outlook Uganda 1989, 1992.23 Switched to CPI from World Economic Outlook Venezuela, RB All years Updated CPI from the National Statistical Office Note: (1) The decimal year notation is used when data are collected over two calendar years. For these countries, a weighted average of the annual CPI series is used, where the weights are based on the data collection. See footnote 3 in Atamanov et al. (2018) and Lakner et al. (2018) for details. 4. Changes to national accounts data The baseline source of national accounts data (per capita GDP and per capita household final consumption expenditure, HFCE) has not been updated from the April 2018 vintage of PovcalNet. It remains the December 2017 version of the World Bank’s World Development Indicators (WDI). A detailed technical note to be published on the PovcalNet website will offer a more comprehensive explanation and documentation of the alternative sources used when WDI data are missing. 6 When more recent national accounts data were needed (e.g. for surveys in 2017), these years were added from the July 2018 vintage of WDI. For Indonesia and the West Bank and Gaza, the national accounts data was chained from 2016 to 2017 due to revision of the series in 2016 or later. For the Maldives, the entire series was updated to the July 2018 version due to large revisions of the national accounts series in the early 2000s. Given the detailed work on revising the line-up procedure (see section 6.2), India’s national accounts data were also updated to the July 2018 version. 5. Changes to population and survey coverage data The baseline source of population data remains the December 2017 version of the WDI, as in the April 2018 vintage of PovcalNet. The total world population has been revised slightly upwards because of four distinct revisions: 1. The following economies have been added to PovcalNet: Andorra, Curacao, Gibraltar, Isle of Man, Nauru, Sint Maarten (Dutch part), St. Martin (French part), Turks and Caicos Islands, British Virgin Islands. Their combined population was 0.49 million people in 2015. 2. For Eritrea, where WDI does not report population in recent years, population estimates from the United Nations World Population Prospects were added for 2012-2017. 3. In the case of Taiwan, China, which was previously missing, population data was added from the National Statistics Republic of China (Taiwan, China). 4. For Kuwait, interpolations have been made between 1991 and 1995, where WDI data were missing. This affects the 1993 line-up. Population survey coverage has been updated. The criteria for estimating survey population coverage is whether at least one survey used in the reference year estimate was conducted within two years of the reference year. 7 6. Revisions to India’s poverty estimates 6.1. CPI revisions Urban and rural CPIs for India have been revised to reflect the most recent data from the Indian Labour Bureau. The revisions primarily impact Indian poverty estimates in the 1980s (surveys in 1983 and 1987.5), but small changes have also been made to the 1993.5 survey and the 2004.5 survey. The biggest change occurred for the 1987.5 rural (urban) survey mean, which was adjusted downwards by 4.8% (6.4%). These revisions have impacts on Indian poverty numbers, as shown in Table 4, and due to India’s size, also on global poverty numbers. For example, the change in the 2004.5 CPI leads to small revisions in the 1996, 1999, 2002, 2005 and 2008 line-ups. For more information on the Indian CPIs, see Lakner et al. (2018). Table 4. Revisions of India CPIs: Comparison of poverty headcount ratio (in %, at $1.90) Urban Rural National Year Apr 2018 Sep 2018 Apr 2018 Sep 2018 Apr 2018 Sep 2018 1983 34.2 36.2 60.0 60.6 53.9 54.8 1987.5 31.0 35.3 49.3 53.5 44.8 49.0 1993.5 29.8 29.7 51.6 51.6 45.9 45.9 2004.5 25.5 25.4 43.4 43.4 38.2 38.2 Source: PovcalNet Note: Survey-year estimates for India not listed in the table remain unchanged. 6.2. Revisions to line-up after 2011.5 The latest survey with official poverty estimates for India was conducted in 2011-12, more than three years before the most recent reference year, 2015. The usual methodology for lining-up countries to the reference year is based on two assumptions: the survey mean grows at the same rate as HFCE or GDP per capita, and there is no other change in the distribution.4 These assumptions may be reasonable when adjusting over a short period of time, but they become 4 See Jolliffe et al. (2015, Box 6.4) for a general description of PovcalNet’s lining up procedure. The forthcoming technical note on the sources of national accounts data will provide a more detailed documentation. 8 problematic as the distance between the survey year and the line-up year increases (Jolliffe et al. 2015). With the usual approach, and with an HFCE growth rate of 21% in India from 2011-12 to 2015, the welfare aggregate for all households in the 2011-12 survey would be given a growth rate of 21%, and poverty in 2015 would be estimated based on this adjusted welfare vector. Given India’s importance for the global poverty rate, and the availability of a newer survey (albeit without a full consumption aggregate, see below), it was felt that this extrapolation method needed to be cross- validated. For this reason, the 2015 poverty estimate for India is based on a new method to estimate the growth rate in HFCE. The new method utilizes a nationally representative survey conducted in 2014-2015 that has similar socioeconomic and demographic information as the 2011-2012 survey, but does not have a full consumption aggregate that can be used for poverty estimation. The 2014- 2015 survey contains information on several household characteristics that are also present in past survey rounds and that can be used to predict per capita consumption. These common characteristics include household age, size, caste, religion, a few labor market variables, and expenditures on miscellaneous services, recreation and transport. Given the unique situation of having essentially the same socioeconomic and demographic data at two points in time, Newhouse and Vyas (2018) use a survey-to-survey imputation method to estimate poverty in 2014-2015.5 The method first estimates the relationship between per capita household consumption and household characteristics using the data from 2004-2005, 2009-2010, and 2011-2012, which have the full consumption questions as well as the variables used in the model. In a second step, the estimated relationship is applied to the 2014-2015 data to predict household consumption and poverty status. PovcalNet uses the poverty rates at $1.90 estimated by Newhouse and Vyas (2018) (10.0% for urban and 16.8% for rural areas) to calibrate the growth rate in survey mean consumption between 2011.5 and 2014.5. The fraction of growth from national accounts that is passed through to growth 5 Newhouse and Vyas (2018) follow the general survey-to-survey imputation technique introduced by Elbers et al. (2003). For the estimation program used (sae command in Stata) and the associated documentation, see Nguyen et al. (2018a, 2018b). 9 in the survey mean implied by this procedure is 55.9% for urban India and 73.3% for rural India.6 It is important to stress that PovcalNet still assumes distribution-neutral growth, but relaxes the assumption that the growth in HFCE per capita is fully transmitted to the survey mean. With this approach, the total growth rate in the survey mean between 2011.5 and 2014.5 is 9.6% in urban India and 12.6% in rural India. This growth rate is distributed to the annual intervals (2012-2014), which are needed for the intermediate line-ups, in proportion to the growth in HFCE observed in national accounts.7 The new method used for India marks the first time that PovcalNet uses inputs from a survey-to- survey imputation method. In the coming years, when countries do not have surveys with full consumption modules, but have other smaller surveys with partial coverage, similar methods may be applied to obtain more timely poverty estimates. Needless to say, household surveys with full consumption modules are undoubtedly the preferred approach, and only in exceptional cases will imputation approaches be relied upon. Table 5 summarizes the poverty estimates for the reference years that have been affected by this revision, for urban and rural India, as well as nationally. For 2015, 9.5% (15.3%) of the population is poor in urban (rural) areas. These are slightly different from the estimates from Newhouse and Vyas (2018) (10.0% and 16.8%, respectively), since HFCE growth rates have been used to line up the estimates from 2014.5 to 2015.8 The 2012-2014 reference year estimates change as well, because the growth rate from 2011.5 to 2012, 2013 and 2014 have been revised with the pass- through factors mentioned above. The estimates published in April 2018 assumed a higher growth in HFCE (a pass-through factor of 100%); assuming now a lower growth rate implies higher poverty rates. 6 Earlier projections had used a pass-through of 57% (for both urban and rural areas) which is based on the observed historical relationship between the survey and national accounts growth rates (Jolliffe et al., 2015, chapter 1, footnote 14; Ravallion, 2003). 7 −2011.5 This is the exact formula used: = 2011.5 + ∗ (2014.5 − 2011.5 ), where refers to 2014.5 −2011.5 the survey mean, and refyear refers to the reference year in question, here 2012, 2013 or 2014. The HFCE data have been updated to the June 2018 vintage of the WDI. 8 A passthrough rate has also been applied to the growth in HFCE from 2014.5 to 2015. Since 2014.5 is constructed as the average of 2014 and 2015 , and since 2014 and 2014.5 are determined using the method above, 2015 is determined as the residual: 2015 = 2 ∗ 2014.5 − 2014 . This implies a passthrough rate from 2014.5 to 2015 of 76.3% for rural India and 59.8% for urban India. 10 The 2010 and 2011 reference year estimates also change. These estimates are based on an interpolation of the 2009.5 and 2011.5 surveys (see Data Appendix of World Bank (2018) for details). While the 2011.5 survey-year estimate is unchanged, the growth rate between 2011 and 2012, and hence also the growth between 2011 and 2011.5 is revised. This causes small changes to these earlier reference year estimates.9 Table 5. Changes in India reference year estimates: Comparison of poverty rates (in %, at $1.90) Year Urban Rural National Apr 2018 Sep 2018 Apr 2018 Sep 2018 Apr 2018 Sep 2018 2010 17.9 17.8 32.5 32.4 28.0 27.9 2011 14.3 14.0 26.1 25.7 22.4 22.0 2012 12.8 13.0 23.2 23.6 19.9 20.3 2013 10.4 11.7 19.3 20.6 16.5 17.8 2015 9.5 15.3 13.4 Source: PovcalNet 9 The 2011 has been revised for the same reason mentioned in the previous footnote (it is determined as the residual of 2011.5 and 2012 ) . Since the growth between 2011.5 and 2012 is revised downwards, the growth between 2011 and 2011.5 is also revised downwards. This lower growth rate implies that the 2011 survey mean based 2011 on extrapolating the 2011.5 vector backwards is higher ( 2011 = 2011.5 ∗ ), and consequently that poverty 2011.5 is lower. This also applies to the 2010 reference year estimate, since the 2011.5 survey is still used there. 11 7. Country-years added 24 new country-years have been added to PovcalNet. These surveys are listed in Table 6. Table 6. New country-years added Economy Years Survey name Bhutan 2017 BLSS: Living Standards Survey China 2015 China National Integrated Household Survey Gabon 2017 EGEP: Enquête Gabonaise pour l'Evaluation et le Suivi de la Pauvreté Indonesia 2017 SUSENAS: National Socio-Economic Survey Ireland 2015 EU-SILC Italy 2015 EU-SILC Kenya 2015.67 IHBS: Integrated Household Budget Survey Kosovo 2016 HBS: Household Budget Survey Luxembourg 2015 EU-SILC Macedonia 2009 EU-SILC Malaysia 2011, 2013, 2015.33 HIS: Household Income Survey Malta 2015 EU-SILC Morocco 2013.5 ENCDM: Enquete Nationale sur la Consommation et les Dépense des Ménages Namibia 2015.27 NHIES: Namibia Household Income and Expenditure Survey Pakistan 2015.5 PSLM: Pakistan Social and Living Standards Measurement Survey Poland 2016 HBS: Household Budget Survey Switzerland 2015 EU-SILC Thailand 2014, 2015 SES: Household Socio-Economic Survey Uganda 2016.5 UNHS: Uganda National Household Survey Vietnam 2016 VHLSS: Vietnam Household Living Standards Survey West Bank and Gaza 2016.75 PECS: Palestinian Expenditure and Consumption Survey 8. Estimating shared prosperity in China The World Bank’s poverty estimates for China are based on tabulated data provided by China’s National Bureau of Statistics. For example, the 2015 estimate is based on 20 points on the urban and rural Lorenz curves. To estimate urban and rural poverty rates, and other distributional statistics, PovcalNet fits parametric Lorenz curves to these grouped data (see the PovcalNet website and the background papers for further details). In addition, PovcalNet makes an adjustment 12 for spatial price differences between urban and rural China, and uses the urban and rural populations from the WDI. PovcalNet reports distributional statistics, including the average consumption of the bottom 40 percent, separately for urban and rural China. However, the World Bank’s Shared P rosperity measure (the growth in average income or consumption of the poorest 40 percent) is defined nationally. Shared Prosperity can be obtained from PovcalNet by using the national poverty gap with the appropriate poverty line. For China, PovcalNet reports the national poverty headcount, as well as the poverty gap, for any poverty line. By rearranging the formula for the poverty gap, it can be shown that the mean consumption of the poor is given by ̅ = × (1 − ) where is the poverty line, the poverty gap, and the poverty headcount ratio. Therefore, the mean consumption of the bottom 40 percent can be found by setting such that the (national) headcount = 0.4. In other words, = 40 , the national 40th percentile. In practice, this involves iterating over poverty lines in PovcalNet until the national = 0.4. The national 40th percentiles are $5.873 and $6.935 (per capita, per day) in 2013 and 2015, respectively. Hence, the mean of the bottom 40 percent is $3.908 and $4.653 in 2013 and 2015, respectively. This implies an annual growth rate in the mean of the bottom 40 percent of 9.11% over this period, which is the Shared Prosperity estimate reported in Chapter 2 of World Bank (2018). As noted above, these results are approximate (e.g. based on 20 points for the urban and rural distributions) and may therefore differ from calculations that use the underlying micro data directly. 9. Other changes The country name for Swaziland was changed to Eswatini. 13 References Atamanov, Aziz, Joao Pedro Azevedo, R. Andres Castaneda Aguilar, Shaohua Chen, Paul A. Corral Rodas, Reno Dewina, Carolina Diaz-Bonilla, Dean M. Jolliffe, Christoph Lakner, Kihoon Lee, Daniel Gerszon Mahler, Jose Montes, Rose Mungai, Minh C. Nguyen, Espen Beer Prydz, Prem Sangraula, Kinnon Scott, Ayago Esmubancha Wambile, Judy Yang, and Qinghua Zhao. 2018. “April 2018 PovcalNet update: What’s new.” Global Poverty Monitoring Technical Note 1. Washington, DC: World Bank. Elbers, Chris, Jean O. Lanjouw, and Peter Lanjouw. 2003. “Micro-Level Estimation of Poverty and Inequality.” Econometrica. 71(1): 355-364. Fatima, Freeha and Nobuo Yoshida. 2018. “Revisiting the poverty trend in Rwanda: 2010/11 to 2013/14.” Policy Research Working Paper 8585. Washington, DC: World Bank. Jolliffe, Dean, Peter Lanjouw, Shaohua Chen, Aart Kraay, Christian Meyer, Mario Negre, Espen Beer Prydz, Renos Vakis, and Kyla Wethli. 2015. A Measured Approach to Ending Poverty and Boosting Shared Prosperity: Concepts, Data, and the Twin Goals. Policy Research Report. Washington, DC: World Bank. Lakner, Christoph, Daniel Gerszon Mahler, Minh C. Nguyen, Joao Pedro Azevedo, Shaohua Chen, Dean M. Jolliffe, Espen Beer Prydz and Prem Sangraula. 2018. “Consumer Price Indices used in Global Poverty Measurement.” Global Poverty Monitoring Technical Note 4. Washington, DC: World Bank. Milanovic, Branko. 1998. Income, Inequality, and Poverty during the Transition from Planned to Market Economy. World Bank Regional and Sectoral Studies. Washington, DC: World Bank Newhouse, David and Pallavi Vyas. 2018. “Nowcasting poverty in India for 2014-15: A survey to survey imputation approach”. Global Poverty Monitoring Technical Note 6. Washington, DC: World Bank. Nguyen, Minh C., Corral Rodas, Paul A., Azevedo, Joao Pedro, Zhao, Qinghua. 2018a. “SAE: Stata module to obtain ELL unit area small area estimates,” Statistical Software Components S458495, Boston College Department of Economics. Nguyen, Minh C, Corral Rodas, Paul A., Azevedo, Joao Pedro., Zhao, Qinghua. 2018b. “Small Area Estimation: An extended ELL approach.” Mimeo, World Bank. Ravallion, Martin. 2003. “Measuring Aggregate Welfare in Developing Countries: How Well Do National Accounts and Surveys Agree?” The Review of Economics and Statistics 85 (3): 645- 52. World Bank. 2018. Poverty and Shared Prosperity 2018: Completing the Poverty Puzzle. Washington, DC: World Bank. 14 Appendix Table A.1 lists the source of CPI used for each country-year reported in PovcalNet. The columns in the table are defined as follows: • Code: The 3-letter country code used by the World Bank: https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country- and-lending-groups • Country name: Name of country • Year(s): Welfare reporting year, i.e. the year for which the welfare has been reported. If the survey collects income for the previous year, it is the year prior to the survey. This is identical to the year variable used in PovcalNet. • CPI period: Common time period to which the welfare aggregates in the survey have been deflated. The letter Y denotes that the CPI period is identical to the year column. When the welfare aggregate has been deflated to a particular month within the welfare reporting year, the month is indicated by a number between 1 and 12, preceded by an M, and similarly with a Q for quarters. The letter W indicates that a weighted CPI is used, as described in equation 1 in the main text. • CPI source: Source of the deflator used. The source is given by the abbreviation, the frequency of the CPI, and the vintage; e.g. IFS-M-201712 denotes the monthly IFS database version December 2017. For country-specific deflators, the description is given in the text or further details are available upon request. 15 Table A.1. Source of temporal deflator used in PovcalNet Code Country name Year(s) CPI period CPI source AGO Angola All W IFS-M-201712 ALB Albania All Y IFS-M-201712 1986-87 Y NSO 1991-2002 M9 NSO ARG Argentina 2003-06 M7-M12 NSO 2007-14 M7-M12 Private estimates 2016- M7-M12 NSO ARM Armenia All Y IFS-M-201712 AUS Australia All Y IFS-A-201712 AUT Austria All Y IFS-M-201712 AZE Azerbaijan All Y IFS-M-201712 BDI Burundi All Y or W IFS-M-201712 BEL Belgium All Y IFS-M-201712 BEN Benin All Y or W IFS-M-201712 BFA Burkina Faso All Y or W IFS-M-201712 1983-88 W WEO-A-201804 BGD Bangladesh 1991- Y or W Survey 1989 Y IFS-A-201712 BGR Bulgaria 1992- Y IFS-M-201712 2001-2004 Y WEO-A-201804 BIH Bosnia and Herzegovina 2007- Y IFS-M-201712 BLR Belarus All Y IFS-M-201712 BLZ Belize All Y IFS-A-201712 1990 W IFS-M-201712 1992, 1997, 2000-02, M11 IFS-M-201712 BOL Bolivia 2005 M10 IFS-M-201712 Rest BRA Brazil All M9 IFS-M-201712 2003 Q2-Q3 IFS-Q-201712 BTN Bhutan 2007- Y IFS-M-201712 BWA Botswana All W IFS-M-201712 CAF Central African Republic All Y or W IFS-M-201712 CAN Canada All Y IFS-M-201712 CHE Switzerland All Y IFS-M-201712 1987 Y ILO-M-201804 CHL Chile 1990-2006 M11 ILO-M-201804 2009- M11 IFS-M-201712 CHN China – Rural All Y NSO CHN China – Urban All Y NSO CIV Cote d'Ivoire All Y or W IFS-M-201712 CMR Cameroon All Y IFS-M-201712 COD Congo, DR All W IFS-M-201712 COG Congo, Republic of All Y IFS-M-201712 16 1988 Y IFS-M-201712 1989-2011 M11 IFS-M-201712 COL Colombia 2012 M9 IFS-M-201712 2013-2015 M11 IFS-M-201712 2016- M8 IFS-M-201712 COM Comoros All Y IFS-M-201712 CPV Cabo Verde All W IFS-M-201712 1981-1989 Y IFS-M-201712 1990-2014 M7 IFS-M-201712 CRI Costa Rica 2015 M5 IFS-M-201712 2016- M7 IFS-M-201712 CYP Cyprus All Y IFS-M-201712 CZE Czech Republic All Y IFS-M-201712 DEU Germany All Y IFS-M-201712 DJI Djibouti All Y IFS-M-201712 DNK Denmark All Y IFS-M-201712 1986-1989 Y IFS-M-201712 1992 M6 IFS-M-201712 DOM Dominican Republic 1996 M2 IFS-M-201712 1997 M4 IFS-M-201712 2000- M9 IFS-M-201712 DZA Algeria All Y or W IFS-M-201712 1987 Y IFS-M-201712 1994 M6-M10 IFS-M-201712 1995 M11 IFS-M-201712 ECU Ecuador 1998 M6 IFS-M-201712 1999 (prev. year) M10-M9 IFS-M-201712 2000- M11 IFS-M-201712 EGY Egypt, Arab Republic of All Y or W IFS-M-201712 ESP Spain All Y IFS-M-201712 EST Estonia All Y IFS-M-201712 ETH Ethiopia All W IFS-M-201712 FIN Finland All Y IFS-M-201712 FJI Fiji All W IFS-M-201712 FRA France All Y IFS-M-201712 FSM Micronesia, FS All Y WEO-A-201804 GAB Gabon All Y IFS-M-201712 GBR United Kingdom All Y IFS-M-201712 GEO Georgia All Y IFS-M-201712 1987-1998 W IFS-M-201712 GHA Ghana 2005- W Survey 1991-2002 Y or W WEO-A-201804 GIN Guinea 2007- Y IFS-M-201712 GMB Gambia, The All Y or W IFS-M-201712 GNB Guinea-Bissau All Y IFS-M-201712 GRC Greece All Y IFS-M-201712 17 1986-1989 Y or W IFS-M-201712 1998 M8 IFS-M-201712 GTM Guatemala 2000 M6-M11 IFS-M-201712 2006- M7 IFS-M-201712 1992 W WEO-A-201804 GUY Guyana 1998- Y IFS-M-201712 1986-1989 Y IFS-M-201712 1990-1993 M5 IFS-M-201712 HND Honduras 1994 M9 IFS-M-201712 1995- M5 IFS-M-201712 HRV Croatia All Y IFS-M-201712 HTI Haiti All M5 IFS-M-201712 HUN Hungary All Y IFS-M-201712 1984-1999 Y IFS-M-201712 IDN Indonesia 2000-2007 M2 IFS-M-201712 2008- M3 IFS-M-201712 IND India - Rural All Y NSO IND India – Urban All Y NSO IRL Ireland All Y IFS-M-201712 IRN Iran, Islamic Republic of All Y NSO IRQ Iraq All Y or W NSO ISL Iceland All Y IFS-M-201712 ISR Israel All Y IFS-M-201712 ITA Italy All Y IFS-M-201712 1988 M9 IFS-M-201712 1990-1993 M11-(next year) M3 IFS-M-201712 JAM Jamaica 1996 M5-M8 IFS-M-201712 1999 M6-M8 IFS-M-201712 2002- M6 IFS-M-201712 JOR Jordan All Y or W IFS-M-201712 JPN Japan All Y IFS-M-201712 KAZ Kazakhstan All Y IFS-M-201712 KEN Kenya All Y or W NSO KGZ Kyrgyz Republic All Y IFS-M-201712 KIR Kiribati All Y IFS-A-201712 KOR Korea, Republic of All Y IFS-M-201712 KSV Kosovo All Y IFS-M-201712 1992-1997 W WEO-A-201804 LAO Lao PDR 2002- W Survey LBN Lebanon All W IFS-M-201712 LBR Liberia All Y IFS-M-201712 LCA St. Lucia All Y IFS-M-201712 LKA Sri Lanka All Y or W IFS-M-201712 1986-1994 W WEO-A-201804 LSO Lesotho 2002- Y or W IFS-M-201712 18 LTU Lithuania All Y IFS-M-201712 LUX Luxembourg All Y IFS-M-201712 LVA Latvia All Y IFS-M-201712 MAR Morocco All W IFS-M-201712 MDA Moldova All Y IFS-M-201712 MDG Madagascar All Y or W IFS-M-201712 MDV Maldives All W NSO MEX Mexico All M8 IFS-M-201712 MKD Macedonia, FYR All Y IFS-M-201712 1994 Y IFS-A-201712 MLI Mali 2001- Y or W IFS-M-201712 MMR Myanmar All M1 IFS-M-201712 MNE Montenegro All Y IFS-M-201712 MNG Mongolia All Y or W IFS-M-201712 1996-2002 W WEO-A-201804 MOZ Mozambique 2008- W IFS-M-201712 MRT Mauritania All Y or W IFS-M-201712 MUS Mauritius All Y or W IFS-M-201712 1997 W IFS-M-201712 MWI Malawi 2004- W Survey MYS Malaysia All Y IFS-M-201712 1993 W WEO-A-201804 NAM Namibia 2003- W IFS-M-201712 NER Niger All Y or W IFS-M-201712 NGA Nigeria All Y or W IFS-M-201712 1993 M2 NSO 1998 M6 NSO NIC Nicaragua 2001 M6 IFS-M-201712 2005-2009 M8 IFS-M-201712 2014 M8-M10 IFS-M-201712 NLD Netherlands All Y IFS-M-201712 NOR Norway All Y IFS-M-201712 NPL Nepal All W IFS-M-201712 PAK Pakistan All Y or W IFS-M-201712 1989 Y IFS-M-201712 1991, 2000, 2006 M6 IFS-M-201712 PAN Panama 2016 M5 IFS-M-201712 Rest M7 IFS-M-201712 1985-1994 Y or W IFS-M-201712 1997-2002 M10-M12 IFS-M-201712 PER Peru 2003 M5-M12 IFS-M-201712 2004- Y IFS-M-201712 PHL Philippines All Y IFS-M-201712 PNG Papua New Guinea All Y IFS-A-201712 1985-1987 Y IFS-A-201712 POL Poland 1989- Y IFS-M-201712 19 PRT Portugal All Y IFS-M-201712 1990 M7 IFS-M-201712 1995 M8-M11 IFS-M-201712 1997 (next year) M2 IFS-M-201712 1999 M9 IFS-M-201712 2001 M3 IFS-M-201712 2002 M11 IFS-M-201712 2003 M7 IFS-M-201712 2004 M10 IFS-M-201712 2005 M11 IFS-M-201712 PRY Paraguay 2006 M12 IFS-M-201712 2007 M10 IFS-M-201712 2008 M8 IFS-M-201712 2009 M11 IFS-M-201712 2010-2011 M10 IFS-M-201712 2012 M2 IFS-M-201712 2013-2014 M10 IFS-M-201712 2015 Y IFS-M-201712 2016- M4 IFS-M-201712 PSE West Bank and Gaza All Y IFS-M-201712 1989 Y Milanovic (1999) ROU Romania 1992- y IFS-M-201712 RUS Russian Federation All Y IFS-M-201712 RWA Rwanda All W IFS-M-201712 SDN Sudan All Y IFS-M-201712 SEN Senegal All Y or W IFS-M-201712 SLB Solomon Islands All Y IFS-M-201712 1989-2003 W WEO-A-201804 SLE Sierra Leone 2011- Y IFS-M-201712 1989 Y IFS-M-201712 1991 M10-(next year) M4 IFS-M-201712 SLV El Salvador 1995-1999 Y IFS-M-201712 2000-2007 M12 IFS-M-201712 2008- M11 IFS-M-201712 SRB Serbia All Y IFS-M-201712 SSD South Sudan All Y IFS-M-201712 STP Sao Tome and Principe All Y or W IFS-M-201712 SUR Suriname All Y IFS-M-201712 SVK Slovak Republic All Y IFS-M-201712 SVN Slovenia All Y IFS-M-201712 SWE Sweden All Y IFS-M-201712 SWZ Swaziland All W IFS-M-201712 SYC Seychelles All Y or W IFS-M-201712 SYR Syrian Arab Republic All Y IFS-M-201712 TCD Chad All Y IFS-M-201712 TGO Togo All Y IFS-M-201712 THA Thailand All Y IFS-M-201712 20 1999 Y WEO-A-201804 TJK Tajikistan 2003-2007 Y Survey 2009- Y IFS-M-201712 TKM Turkmenistan All Y WEO-A-201804 2001 Y WEO-A-201804 TLS Timor-Leste 2007- Y IFS-M-201712 TON Tonga All Y IFS-M-201712 TTO Trinidad and Tobago All Y IFS-M-201712 1985 Y IFS-A-201712 TUN Tunisia 1990- Y or W IFS-M-201712 TUR Turkey All Y IFS-M-201712 TUV Tuvalu All Y WEO-A-201804 1991 Y IFS-A-201712 TZA Tanzania 2000- Y or W IFS-M-201712 1989-1992 W WEO-A-201804 UGA Uganda 1996- W IFS-M-201712 UKR Ukraine All Y IFS-M-201712 1981-1989 Y IFS-M-201712 URY Uruguay 1992- (prev. year) M12 IFS-M-201712 USA United States All Y IFS-M-201712 UZB Uzbekistan All Y WEO-A-201804 1981-1989 Y NSO VEN Venezuela 1992- M12 NSO 1992 W WEO-A-201804 VNM Vietnam 1998 W IFS-M-201712 2002- M1 IFS-M-201712 VUT Vanuatu All Y IFS-A-201712 WSM Samoa All Y IFS-M-201712 YEM Yemen, Republic of All Y or W IFS-M-201712 1993-2000, 2008 Y or W IFS-M-201712 ZAF South Africa 2005, 2010- (next year) M6 IFS-M-201712 ZMB Zambia All Y or W IFS-M-201712 ZWE Zimbabwe All Y IFS-M-201712 MLT Malta All Y IFS-M-201712 21