A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA © 2024 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 conclu- sions 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, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. 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. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immu- nities of The World Bank, all of which are specifically reserved. 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. Attribution: Please cite the work as follows: World Bank (2024) “A Fresh Take on Reducing Inequality and Enhancing Mobility in Malaysia”, World Bank, Washington, DC. 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. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA I CONTENTS Acknowledgments��������������������������������������������������������������������������������������������������������������������������������������� 1 Overview�������������������������������������������������������������������������������������������������������������������������������������������������������2 1. Inequality: Concepts, measurement, and why it matters��������������������������������������������������������������� 7 1.1 Why should we care about inequality in Malaysia?��������������������������������������������������������������������������������������������8 1.2 Measuring inequality���������������������������������������������������������������������������������������������������������������������������������������������������� 10 1.3 Mobility�������������������������������������������������������������������������������������������������������������������������������������������������������������������������������13 2. Inequality in Malaysia over the past two decades�������������������������������������������������������������������������� 15 2.1 Explaining inequality trends within the country����������������������������������������������������������������������������������������������25 3. Economic mobility trends������������������������������������������������������������������������������������������������������������������ 34 3.1 Absolute income mobility������������������������������������������������������������������������������������������������������������������������������������������35 3.2 Relative income mobility������������������������������������������������������������������������������������������������������������������������������������������ 42 4. Understanding the drivers of inequality in Malaysia������������������������������������������������������������������� 44 4.1 Inequality framework: Assets�����������������������������������������������������������������������������������������������������������������������������������45 4.2 Inequality framework: Returns to assets in the labor market����������������������������������������������������������������������51 Inequality framework: Shocks����������������������������������������������������������������������������������������������������������������������������������������67 Inequality framework: Taxes and public spending�������������������������������������������������������������������������������������������������73 5. Addressing inequality and promoting mobility: What can be done?����������������������������������������� 75 5.1 Increasing opportunities for all Malaysians�������������������������������������������������������������������������������������������������������� 80 5.2 Strengthening social protection to address the needs at the bottom of the income distribution ���������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������� 84 5.3 Addressing spatial inequalities���������������������������������������������������������������������������������������������������������������������������������87 5.4 Financing inclusive investments���������������������������������������������������������������������������������������������������������������������������� 88 5.5 Broadening the inequality monitoring toolkit�������������������������������������������������������������������������������������������������� 96 References�������������������������������������������������������������������������������������������������������������������������������������������������� 99 Appendix A. Welfare and inequality measurement in Malaysia���������������������������������������������������������������������106 Appendix B. Survey of Malaysians on Perceptions of Inequality and Intergenerational Mobility, 2023�����������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������108 Appendix C. Additional figures and tables for mobility estimates������������������������������������������������������������������110 Appendix D. Definitions and methodology for climate risk analysis������������������������������������������������������������115 Disaster risk���������������������������������������������������������������������������������������������������������������������������������������������������������������������������115 Hazard�������������������������������������������������������������������������������������������������������������������������������������������������������������������������������������116 I A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA LIST OF FIGURES Figure 1.1 Malaysia is on a path to achieve high-income status, but many of its people are not��������8 Figure 1.2 Income generation asset framework to understand inequality������������������������������������������������� 10 Figure 1.3 Opportunities start early in life and compound over time���������������������������������������������������������� 12 Figure 1.4 Correlation between income inequality and inequality of opportunity��������������������������������13 Figure 2.1 Malaysia Gini index long-run trend, 1970–2022������������������������������������������������������������������������������ 16 Figure 2.2 Inequality in Malaysia is higher than in transitional and established high-income countries�����������������������������������������������������������������������������������������������������������������������������������������������������������17 Figure 2.3 Malaysia Gini index, 2004–22������������������������������������������������������������������������������������������������������������������17 Figure 2.4 Inequality trend with different measures, 2004–22����������������������������������������������������������������������� 18 Figure 2.5 Perceptions of income gap 10 years ago among actual income classes��������������������������������� 19 Figure 2.6 Perceptions of income gap today among actual and perceived income classes���������������� 19 Figure 2.7 Absolute versus relative Gini, 2004–22��������������������������������������������������������������������������������������������� 20 Figure 2.8 Matrix of actual and perceived income classes of households������������������������������������������������ 20 Figure 2.9 Share of total wealth held by the richest 10 percent of households (percent)�������������������� 21 Figure 2.10 Labor income share of GDP, 2020������������������������������������������������������������������������������������������������������� 22 Figure 2.11 Aggregate net worth of Malaysia’s billionaires as a share of GDP, 2001–23������������������������� 22 Figure 2.12 Billionaires’ wealth as a share of GDP, 2001, 2010, and 2022������������������������������������������������������23 Figure 2.13 Prosperity gap in Malaysia (high-income status threshold in 2017 $PPP), 2004–22�������� 24 Figure 2.14 Decomposition of the prosperity gap in Malaysia, 2004–14, 2014–19, 2019–22���������������� 24 Figure 2.15 Growth incidence curves, 2004–14, 2014–19, 2019–22�����������������������������������������������������������������25 Figure 2.16 Income shares across the distribution, 2004–22�����������������������������������������������������������������������������25 Figure 2.17 Sources of inequality change, Gini points, 2004–14, 2014–19, 2019–22�������������������������������� 26 Figure 2.18 Labor income share for Malaysia, 2004–22�������������������������������������������������������������������������������������� 26 Figure 2.19 Trend in mean incomes per capita by state, 2004–22������������������������������������������������������������������27 Figure 2.20 Relative wealth index (projection to 100 m populated areas)��������������������������������������������������� 28 Figure 2.21 Income shares by ethnic group, 2004–22����������������������������������������������������������������������������������������� 29 Figure 2.22 Growth incidence curves by ethnicity, 2004–22��������������������������������������������������������������������������� 29 Figure 2.23 Mean income by ethnic group, 2004–22������������������������������������������������������������������������������������������� 30 Figure 2.24 Prosperity gap trends by ethnic group, 2004–22��������������������������������������������������������������������������� 30 Figure 2.25 Mean income by ethnic group and region, 2004–22�������������������������������������������������������������������� 30 Figure B2.1.1 Change in household size across the income distribution, 2004–22��������������������������������������31 Figure B2.1.2 Inequality for different fertility scenarios, 2004, 2014, 2022��������������������������������������������������������31 Figure B2.1.3 Inequality (mean log deviation) within birth cohorts, 2005–20�����������������������������������������������32 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA II Figure B3.1.1 B40/M40/T20 and economic security classes��������������������������������������������������������������������������������36 Figure 3.1 Chronic poverty by ethnic group����������������������������������������������������������������������������������������������������������37 Figure 3.2 Chronic poverty within the Bumiputera��������������������������������������������������������������������������������������������37 Figure 3.3 Absolute income mobility by ethnicity (economically secure class)��������������������������������������38 Figure 3.4 Absolute income mobility by rural/urban and Peninsular Malaysia/East Malaysia (eco- nomically secure class)�������������������������������������������������������������������������������������������������������������������������������39 Figure 3.5 Absolute income mobility by ethnicity and Peninsular Malaysia/East Malaysia (eco- nomically secure class)�������������������������������������������������������������������������������������������������������������������������������39 Figure 3.6 Perceived income mobility (head of household vis-à-vis parents) by perceived income class�������������������������������������������������������������������������������������������������������������������������������������������������������������������40 Figure 3.7 Difference between actual and perceived income mobility for children����������������������������40 Figure 3.8 Short-term relative mobility: number of quintiles moved up or down��������������������������������� 42 Figure 3.9 Persistence at the bottom and top by region and the Bumiputera ethnicity���������������������� 43 Figure 4.1 Inequality of opportunity (at market income)����������������������������������������������������������������������������������45 Figure 4.2 Human Development Index among aspirational countries, 1990–2021��������������������������������45 Figure 4.3 Rates of stunting and low birthweight by state and ethnicity��������������������������������������������������� 46 Figure 4.4 Mean performance in PISA tests by quarter of socioeconomic status����������������������������������47 Figure 4.5 Percentage of children engaged in at-home learning in the past 30 days (May/June 2021), by household income���������������������������������������������������������������������������������������������������������������������47 Figure 4.6 Malaysia-specific subnational Human Capital Index and state GDP per capita��������������� 48 Figure 4.7 Educational attainment of the Bumiputera, 2004–22����������������������������������������������������������������� 49 Figure 4.8 Educational attainment by group and decile, 2022���������������������������������������������������������������������� 49 Figure 4.9 Student performance (UPSR) and educational attainment������������������������������������������������������� 50 Figure 4.10 Income sources as a share of total income by income decile, 2019�����������������������������������������51 Figure 4.11 Gini for employment income per capita, 2004–22�������������������������������������������������������������������������51 Figure 4.12 Growth incidence curve for employment income per capita, 2004–22������������������������������52 Figure 4.13 Sector of employment by sector and decile, 2004, 2014, and 2022����������������������������������������52 Figure 4.14 Output per worker grew the most in sectors where the richest work�����������������������������������53 Figure 4.15 Sector of employment by education level�����������������������������������������������������������������������������������������53 Figure 4.16 Real median employment income by ethnicity, 2004-22����������������������������������������������������������� 54 Figure 4.17 Explained and unexplained differences in distribution of total employment earnings, 2004-22����������������������������������������������������������������������������������������������������������������������������������������������������������� 54 Figure 4.18 Wage income growth by education level, 2010–22������������������������������������������������������������������������55 Figure 4.19 Average skill premium, 2004–22�����������������������������������������������������������������������������������������������������������55 Figure 4.20 Skill premium by per capita household income ventil, 2004–22���������������������������������������������56 Figure 4.21 Returns on education by per capita household income ventil, 2004–22�����������������������������57 Figure 4.22 Activity in household income deciles 1, 5, and 10����������������������������������������������������������������������������58 Figure 4.23 Occupations by skill level in household income deciles 1, 5, and 10���������������������������������������58 Figure 4.24 Informality by income decile, 2009–19�����������������������������������������������������������������������������������������������58 III A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Figure 4.25 Occupations by skill and state����������������������������������������������������������������������������������������������������������������59 Figure 4.26 Top three business environment constraints by size���������������������������������������������������������������������59 Figure 4.27 Share of firms and difficulty to satisfy the demand for high-skilled labor, by firm size��� 60 Figure 4.28 Jobs by skill level, 2010–21������������������������������������������������������������������������������������������������������������������������� 61 Figure 4.29 Skill-related underemployment by age, 2010–21����������������������������������������������������������������������������� 61 Figure 4.30 Skills by sector���������������������������������������������������������������������������������������������������������������������������������������������� 62 Figure 4.31 Skills by ethnicity����������������������������������������������������������������������������������������������������������������������������������������� 62 Figure B4.1.1 Educational attainment for men and women, 2022����������������������������������������������������������������������63 Figure B4.1.2 Male and female labor force participation (by education), 2010–21���������������������������������������63 Figure B4.1.3 Work participation across the lifecycle for men and women (age of the youngest child)���������������������������������������������������������������������������������������������������������������������� 64 Figure B4.1.4 Work participation across the lifecycle for men and women (age of the oldest child)� 64 Figure B4.1.5 Skill level at work by gender, 2022��������������������������������������������������������������������������������������������������������65 Figure B4.1.6 hare of employment by sector and gender, 2022����������������������������������������������������������������������������65 Figure B4.1.7 Oaxaca-Blinder decomposition for men vs. women. Differences in total and hourly earnings, 2010–22�����������������������������������������������������������������������������������������������������������������������������������������65 Figure B4.1.8 Oaxaca-Blinder decomposition for married men vs. married women. Differences in total and hourly earnings, 2010–22��������������������������������������������������������������������������������������������������������65 Figure 4.32 Importance of hard work for economic prospects����������������������������������������������������������������������� 66 Figure 4.33 People find it difficult to improve their economic situation even if they worked hard�� 66 Figure 4.34 Vulnerability over time in Malaysia, 2004–22����������������������������������������������������������������������������������67 Figure 4.35 Vulnerability by group, 2019�������������������������������������������������������������������������������������������������������������������� 68 Figure 4.36 Vulnerability by state, 2019���������������������������������������������������������������������������������������������������������������������� 68 Figure 4.37 Shock’s impacts on mobility in Q1 and Q5���������������������������������������������������������������������������������������� 69 Figure 4.38 Climate hazard risk and relative living standards at the district level��������������������������������������71 Figure 4.39 Days with warm nights exceeding 35°C, 1950–2100����������������������������������������������������������������������72 Figure 4.40 Gini index before and after fiscal policy, 2019 (percentage)�������������������������������������������������������73 Figure 4.41 Taxes and transfers by income decile, 2019 (percentage of market income)����������������������73 Figure 4.42 Change in Gini index due to different fiscal instruments (percentage points)�������������������74 Figure 5.1 Acceptance of income gap by actual and perceived income class�����������������������������������������76 Figure 5.2 Preference for the time taken for the Malaysian government to address inequality in Malaysia, by perceived income class����������������������������������������������������������������������������������������������������77 Figure 5.3 Growth decomposition for Malaysia: (a) 1991–2021 and (b) projections for 2021–50����� 80 Figure 5.4 Returns on a unit of dollar invested������������������������������������������������������������������������������������������������������83 Figure 5.5 Coverage, share and value of social assistance benefits by decile, 2019 and 2022������������85 Figure 5.6 Share of subsidies by decile, 2019����������������������������������������������������������������������������������������������������������85 Figure 5.7 Poverty and inequality reduction (Gini index) and fiscal savings (percent of GDP) under fuel subsidy and social assistance reform scenarios��������������������������������������������������������� 86 Figure 5.8 Education and health spending, 2004–22���������������������������������������������������������������������������������������� 88 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA IV Figure 5.9 Government revenue by source, percent of GDP������������������������������������������������������������������������� 89 Figure 5.10 Government expenditure by type, percent of GDP��������������������������������������������������������������������� 89 Figure 5.11 Inequality reduction (Gini index) and fiscal savings (percent of GDP) under different scenarios of fuel subsidies, value added, and social assistance reforms, relative to baseline GST simulations������������������������������������������������������������������������������������������������������������������� 90 Figure 5.12 Malaysia’s impact on inequality (Gini index) through taxes, transfers, subsidies, and in-kind services: Baseline and reform scenarios������������������������������������������������������������������������������ 91 Figure 5.13 Size of the cash transfer budget and public perception of the reasons for poverty�������� 92 Figure 5.14 Most popular policies for reducing inequality����������������������������������������������������������������������������������93 Figure 5.15 Most popular policies for reducing inequality��������������������������������������������������������������������������������� 94 Figure 5.16 Perception of income tax��������������������������������������������������������������������������������������������������������������������������95 Figure 5.17 Perception of income tax rates among the rich by actual income������������������������������������������95 Figure 5.18 Prosperity gap in Malaysia and EAP (1990–2019)��������������������������������������������������������������������������� 96 Figure 5.19 Statistical Performance Indicators pillars in Malaysia�������������������������������������������������������������������97 Figure A.1 Gini index, 2004–22 (per capita vs. household total income)�������������������������������������������������107 Figure A.2 Gini index, 2004–22 (with and without spatial deflation)��������������������������������������������������������107 Figure C.1 Chronic poverty transitions (population shares)���������������������������������������������������������������������������110 Figure C.2 Chronic poverty by ethnicity and geography����������������������������������������������������������������������������������110 Figure C.3 Absolute income mobility transitions (economically secure class)�����������������������������������������111 Figure C.4 Relative mobility between income quintiles, 2004–07 to 2019–22����������������������������������������111 Figure C.5 Relative mobility between income quintiles, 2004–07 to 2019–22���������������������������������������112 Figure C.6 Inequality within age cohorts—GE 1 (stratum: urb), individual income�������������������������������113 Figure C.7 Inequality within age cohorts—GE (-1) (stratum: urb), income recipients��������������������������114 Figure C.8 Inequality within age cohorts—Gini (stratum: urb), individual income�������������������������������114 Figure D.1 Hazard, risk, vulnerability, and exposure�������������������������������������������������������������������������������������������115 Figure D.2 Exposure categories considered in the analysis and related impact types�������������������������116 Figure D.3 Example impact model�����������������������������������������������������������������������������������������������������������������������������118 Figure D.4 Computation of annual expected impact of natural hazards in geospatial analytics�����119 Figure D.5 Relative wealth index (district level mean)��������������������������������������������������������������������������������������120 Figure D.6 Population exposure to river floods in relation to living standards (RWI)�������������������������120 Figure D.7 Population exposure to coastal floods in relation to living standards (RWI)��������������������120 Figure D.8 Population exposure to landslides in relation to living standards (RWI)�����������������������������121 Figure D.9 Population exposure to head stress in relation to living standards (RWI)���������������������������121 Figure D.10 Returns to education by quantiles of the labor income distribution, 2004-22�����������������122 V A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA LIST OF TABLES Table 2.1 Share of inequality attributable to the intergroup component (%)���������������������������������������� 28 Table 4.1 Income sources as a share of total income, 2004–22���������������������������������������������������������������������51 Table 4.2 Expected annual impact from a probabilistic analysis of flood risk to population and- built-up area����������������������������������������������������������������������������������������������������������������������������������������������������72 Table 5.1 Policy measures for reducing inequality and promoting economic mobility in Malaysia ������������������������������������������������������������������������������������������������������������������������������������������������������������������������������77 Table B.1 Sample distribution of Survey of Malaysians on Perceptions of Inequality and Intergen- erational Mobility 2023����������������������������������������������������������������������������������������������������������������������������108 Table C.1 Counterfactual data�����������������������������������������������������������������������������������������������������������������������������������112 Table D.1 Available hazard, exposure, and vulnerability components from global data sets����������� 117 Table D.2 Selected risk classification approach by hazard�����������������������������������������������������������������������������119 Table D.3 Climate variables underlying climate projections������������������������������������������������������������������������122 LIST OF BOXES Box 2.1 Household demographics and inequality��������������������������������������������������������������������������������������������31 Box 3.1. Income groups and economic security classes in Malaysia��������������������������������������������������������35 Box 4.1 Gender gaps in the Malaysian labor market��������������������������������������������������������������������������������������63 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA VI ACKNOWLEDGEMENTS A Fresh Take on Reducing In- Tax Division and the Inland Revenue Board of equality and Enhancing Mo- Malaysia team, in particular Encik Shaharrudy bin bility in Malaysia is a product Othman, Roshida binti Daud, and Aiman Hakim, of The World Bank’s Pover- whose collaborative efforts and dedication have ty and Equity Global Prac- been crucial in gathering and preparing the data tice team in the Malaysia for the analysis presented in this report. Hub office, in close collab- oration with the Ministry of Economy Malaysia This report was prepared by a core World Bank and Department of Statistics Malaysia (DOSM). team led by Laura Rodríguez, Matthew Wai-Poi and Ririn Salwa Purnamasari, and consisted of This report would not have been possible with- Nishant Yonzan, Christoph Lakner, Kenneth Sim- out the support and guidance of YB Dato’ Seri ler and Alyssa Farha Binti Jasmin. Background Rafizi Ramli, Minister of Economy and YB Dato’ work on mobility was conducted by Peter Lan- Sri Mustapa bin Mohamed, Former Minister at jouw and Gerton Rongen. Background analy- Prime Minister’s Department (Economy). In- sis of the perceptions survey was conducted by valuable contributions to this report were pro- Muhammed Bin Abdul Khalid, Zouhair Rosli and vided by the Ministry of Economy team, under Ririn Salwa Purnamasari. Important inputs to the the leadership and guidance of Dato’ Nor Azmie report were received from Yew Keat Chong, Mat- Diron and Datuk Dr. Zunika Mohamed, specifi- thew Dornan, Ya Shin Wa, Nadia Esham, Lars M. cally Mr. Azlan Abdul Rashid, Mr. Moktar Idham Sondergaard, Daniel Halim, Apurva Sanghi, Mei Musa, Ms. Ashikin Abdul Razak, Dato’ Herman Ling Tan, Mattia Amadio, Veronica Montalva Tall- Abdul Hamid, Mr. Fairul Rafieq Abdullah, Dr. Rid- edo and Shahira Zaireen Bt Johan Arief Jothi. zuan Kushairi Mohd Ramli, Dr. Irwan Wahyudi Ibrahim, Mr. Mohd Zainal Othman, Ms. Siti Arfah Excellent comments were received from peer Kamaruzaman, Ms. Malisa Mat Noor, Ms. Nur Ain reviewers Maria Davalos, Hwok-Aun Lee, Rich- Muhammad Yusuf, Ms. Tan Fung Ling, and the ard Record, Liliana Do Couto Sousa, Federico Equity Development Division. Their cooperative Gil Sander, Christopher Hoy, and from Gonzalo spirit, insights, active participation in series of dis- J. Varela, also from various Divisions in the Min- cussions and review process, as well as efforts in istry of Economy (Environmental and Natural coordinating feedback from various stakeholders Resources, Macroeconomics, Security and Pub- have been instrumental in the successful comple- lic Order, Human Capital Development, Agri- tion of this quality and relevant report. culture, Knowledge Economy Division), DOSM, Ministry of Finance, Ministry of Health, Ministry The team is indebted to the Malaysian Depart- of Education, Ministry of Human Resource, and ment of Statistics (DOSM), under the guidance Ministry of Women, Family And Community of Dato’ Seri Dr. Mohd Uzir Mahidin, for making Development, as well as from participants in the available promptly the Household Income, Ex- three workshops conducted for this report. This penditure and Basic Amenities Survey (HIES/BA) report benefitted greatly from these comments. and for conducting the Survey of Malaysians on Perceptions of Inequality and Intergenerational The report was edited by Steven B. Kennedy. Mobility in partnership with the World Bank spe- Graphic design and layout of the report was cifically for this report, as well as to the Ministry of created by Mikael Jonathan Bima, with support Finance (MOF), under the guidance of Dato’ Che on graphs by Goldy Dharmawan. Substantial Nazli binti Jaapar, for providing access to personal administrative support was provided by Mini- income tax data to be used in this report. Special sha Mandeepal and Dyah Kelasworo Nugraheni. thanks go to the DOSM team, in particular Ms. Nazaria Baharudin, Ms. Siti Asiah Ahmad, Mr. The report was produced under the overall guid- Azmi Ali, Ms. Tassha Hilda Adnan, Mr. Abul Ha- ance of Rinku Murgai and Benu Bidani. Strategic fidz Abdul Hamid, Mr. Muhammad Amjad Mohd guidance and key comments were provided by Jailani, Ms. Nor Anees Asyiqin Md Disa, Mr. Nik Yasuhiko Matsuda, Ndiame Diop, Hassan Zaman, Muhammad Hafiz Azlan, and Price, Income and Lalita Moorty, Zafer Mustafaoglu, and Judith Expenditure Statistics Division, and to the MOF Green. 1 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA OVERVIEW A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 2 F or the first time, this re- Malaysia has an admirable record of econom- port looks comprehen- ic growth and poverty reduction. More is sively at the evolution of needed to make the economy more inclusive, inequality and mobility achieve the country’s economic and human in Malaysia over the last development objectives, and improve Malay- two decades and into the sians’ well-being. Income inequality1 is much post-Covid era, as well as lower today than it was 50 years ago, nonetheless what drives these trends. This report includes progress on reducing it appears to have stalled income inequality trends since the Covid-19 over the past two decades, and inequality in Ma- pandemic, utilizing the most recent household laysia is higher than it is in high-income countries survey data from the 2022 Household Income, (HICs) as well as in countries that became HIC Expenditure and Basic Amenities Survey (HIES/ recently. Unless growth is shared more evenly, in BA). These data are included as part of a compre- 2028-2030, when Malaysia is projected to pass hensive analysis of inequality that goes beyond the HIC threshold, less than half of all Malaysians the Gini coefficient to understand how inequality will have incomes above the threshold. Greater has evolved over the last two decades in Malaysia. emphasis on reducing inequality—especially by What is also new in this report is, for the first time, enhancing equality of opportunity through ex- a measure of the related but distinct concept of panded access to high-quality education, health income mobility – how much individuals move services, and productive employment—can also up or down the income distribution over time. In help reinvigorate growth. addition, while previous works have focused on ethnic gaps or on regional gaps, this report em- Reducing inequality is a priority to Malay- phasizes that a great extent of inequality occurs sians. Reduction of inequality figures promi- within these groups, and highlights the intersec- nently in Malaysia’s major planning and strategy tion of ethnicity and location as a salient mark- documents, such as the 12 Malaysia Plan and its er of inequality. Finally, whereas other research midterm review and in the MADANI Econom- presents inequality trends without an analysis ic Framework –the most recent strategic vision of what drives them, or focuses just on a single document—which elevates inclusive growth and driver of inequality, this report examines a range equality of opportunities as key missions. Malay- of them, including: access to health, education sian citizens also care about inequality. Accord- and other public services, the quality of those ing to a new survey conducted for this report, 70 services, access to employment opportunities percent of respondents indicated that the gap and the returns to education, the role of shocks between the rich and poor is wide or very wide and the role of fiscal policy. In using a lifecycle and 63 percent reported that reducing inequality approach to understand these drivers, the report should be an urgent task for government. shows how gaps in income gaps today are the result of gaps in opportunities earlier in life, and Addressing inequality requires overcoming moreover how these disparities compound on the barriers that limit economic mobility. each other over the lifecycle. Malaysia has been successful at fostering abso- lute economic mobility: Nearly all Malaysians In addition, the report asks what can be done have more education, better health, a higher-pay- to make Malaysia a more equal country as it ing job, and greater material comfort than their makes the transition to a high-income na- parents or grandparents did at a similar age. In tion. After the analysis of trends and diagnostics, contrast, relative mobility—the extent to which the report turns to policies. First, it reexamines people move up or down the economic ladder policy priorities for reducing inequality in the compared with other people over time—is com- context of today’s Malaysia. Second, it outlines a paratively low. More than half of people born in series of fiscal reforms, including new taxes and the poorest 20 percent of the income distribu- better uses of existing spending, that would both tion tend to remain at the bottom as adults, and reduce the post-pandemic fiscal deficit while nearly two-thirds of those born in the richest 20 expanding spending on equity- and growth- percent are likely to remain there. The dynamics enhancing policies. Finally, using a bespoke survey of mobility are just as important as the snapshot conducted for this report, it looks at Malaysians’ of inequality of outcomes at a particular point in perceptions of inequality and mobility, how big time—not least because a given level of inequali- they think income gaps are and how concerned ty is easier to tolerate if people believe they have it makes them, and what policies they think are a reasonable chance of getting ahead through most important in addressing inequality. hard work and effort no matter where they start. 1. Much less is known about the status and trends in wealth inequality in Malaysia. 3 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Economic differences within ethnic groups possess and how effectively they use those skills and locations are far more important than on the job. Much of the earnings inequality ob- differences between them. Ethnicity has long served in the labor market can be traced back to been the dominant thread of inequality dialogue inequalities early in life, from mothers’ antenatal and policy in Malaysia. Nationally, the average care to early childhood nutrition to the quality income of Bumiputera—the group that makes of schooling. Redressing these inequalities later up the majority of the population—is still lower in life is costly and difficult; early intervention is than that of Chinese Malaysians and Indian Ma- therefore essential. laysians, although the gap has slowly narrowed over the last 20 years. Averages masks differences Declining skills premiums in recent years, within groups, however. In 2022, only 13 percent coupled with weak growth in high-skill job of total income inequality in Malaysia was ex- creation, may be contributing to the stagna- plained by differences in average income across tion in inequality reduction. Access to higher ethnicities; 87 percent reflected differences with- education has expanded rapidly in Malaysia, but in each group2. A very similar result holds when the returns to higher education are lower for comparing income differences across and within workers in poorer households, even after con- states: 86 percent of total inequality reflects dif- trolling for other individual characteristics. As ferences within rather than between states. in other countries, the mismatch between what students learn and the skills that employers are The average income differences between seeking is large. In addition, educational expan- state and ethnic groups together explain 20 sion and the slow growth of high-skills jobs has percent of total income inequality. Income led to high and rising rates of skills-related under- gains among Bumiputera in East Malaysia (Sabah, employment, especially among recent graduates. Sarawak, and Labuan) have been much lower The premium for higher education relative to than those of Bumiputera on the peninsula. In- completing only secondary schooling is smaller come inequality between the peninsula and East when the job obtained does not make full use of Malaysia is compounded by, and partly a result a worker’s skills. People with the best education of, inequality in access to good-quality health and most in-demand skills continue to achieve and education services. Spatial inequality is per- high incomes, but higher education and better sistent, with economic mobility lowest among jobs are not benefiting the majority as much as the Bumiputera in East Malaysia. But although they used to and are therefore not contributing to spatial and ethnic disparities matter—and the inequality reduction as much as they could. intersection of them matters even more—the differences between poor and rich Bumiputera What can Malaysia do to address inequality are much larger than the differences between the and promote economic mobility? Table O.1 average Bumiputera and the average non-Bumi- presents policy measures in 5 areas to reduce putera. inequality and promote economic mobility by addressing the challenges identified throughout Labor earnings are the principal source of in- the report. In addition, there is a sharper focus on come for most Malaysians and a key part of how to pay for the package of policies identified trends in income inequality. How much peo- and how to communicate to the public why this ple earn depends in large part on what skills they package is needed and how it will benefit them. 2. This was very similar before the COVID-19 pandemic. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 4 TA B L E O .1 POLICY ME ASURES FOR REDUCING INEQUALIT Y AND PROMOTING E C O N O M I C M O B I L I T Y I N M A L AY S I A Theme Policy measure Enhancing • Increase productivity and make better use of underutilized labor. opportunities • Improve the quality of education and skills development in areas and among households that lag the national average. • Increase investment in broad-based human capital development early in the lifecycle. Strengthening social • Increase spending on social protection. protection • Improve targeting across the income distribution to increase spending efficiency. • Reduce fragmentation and improve administrative efficiency. Financing inclusive • Increase investments in education and health for accelerated and inclusive growth. investments • Create the fiscal space for greater equity-improving investments. • Communicate the benefits of the package of equity-enhancing policies to the public. Addressing spatial • Use place-based and place-sensitive policies to address spatial inequalities. inequalities Monitoring • Track inequality using a suite of measures. inequality • Systematically track economic mobility as well as inequality. A crucial element of reducing inequality is concentrate the benefits more on poorer house- enhancing equality of opportunity by invest- holds would increase the cost-effectiveness of ing in foundational human capital, such as the system. At the same time, larger budgets are health and education, early in the lifecycle. needed to bring spending levels in line with the Closing access gaps among poor and rich house- UMIC average. However, the inequality-reducing holds as well as across regions, economic classes, impact can be enhanced further—even within and ethnic groups is necessary but not sufficient current budget levels—by redirecting much of to reduce inequality; the much larger gaps in qual- the current spending on untargeted subsidies to ity also need to be eliminated. Early interventions targeted social assistance directed to those that are more cost-effective than later-life remedia- need it the most. tion efforts and help to balance the playing field and increase economic mobility. Investments Broad-based policies can help lift poorer on foundational human capital through policies households closer to the rest of the country, which ensure that more children, especially chil- directly benefiting poorer Bumiputera, who dren from disadvantaged backgrounds, reach and are disproportionately represented among are prepared for tertiary schooling. Such policies the poor. Improvements in educational attain- are both pro-equity and pro-growth, increasing ment helped close the earnings gap between labor income and economic mobility by raising Bumiputera and other workers over the last two productivity. They should be matched by mea- decades. Endowment gaps, primarily in educa- sures to accelerate the creation of high-skills tion, explained 77 percent of the earnings differ- jobs. As spatial inequalities remain significant, ence between Bumiputera and non-Bumiputera place-based and place-sensitive policies will also in 2004–12 but just 58 percent in 2014–22. The be needed. contribution to earnings differences between the two groups that cannot be explained by differenc- Strengthening the social protection system es in worker endowments actually rose between by setting more strategic spending priorities the two periods, accounting for 42 percent of the would directly reduce inequality. Inequality in total earnings gap in 2014-22 compared with 23 Malaysia is high when measured based on house- percent in 2004-12. More research is needed to holds’ market income. More relevant for poli- understand why Bumiputera workers earn less cy, it remains high even after richer households than non-Bumiputera workers with similar levels pay taxes and poorer ones receive transfers, be- of education and therefore what complementary cause the country’s fragmented social assistance policies may be needed. This may reflect, for ex- system reduces inequality less than it does in ample, (unmeasured) differences in educational HICs and other upper-middle-income countries quality, or the fields of study chosen, or soft-skills (UMICs). Consolidating programs to improve ad- acquired, among others. The key objective for any ministrative efficiency and improving targeting to complementary policies that focus on ethnicity is 5 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA to ensure they benefit disadvantaged Bumiputera and the World Bank for this report reveals clear rather than those who are already succeeding. public support for prioritizing interventions that reduce inequality, many of which require in- Additional revenues are needed to finance creased public spending. The public is likely to be investments in human capital and a robust more receptive to tax increases (and subsidy re- social protection system. Increased fiscal movals) if it sees that the funds are being spent ef- space can be created through a phased set of tax fectively to address their concerns, which largely reforms. A first phase could increase indirect con- parallel the policies recommended in this report. sumption taxes, offsetting the burden on the poor by providing a rebate for low-income households. A broader set of indicators is needed to mon- A second phase could involve personal income itor progress in reducing inequality. In addi- tax (PIT) reforms that broaden the base by lower- tion to the static income inequality measures ing the taxable income thresholds, increase col- that are currently collected, more information lections, set and implement a cap on total relief is needed on inequality of opportunity and eco- claimed, and raise the marginal tax rates for high- nomic mobility. One way to obtain this informa- er income brackets. tion would be to add retrospective questions to major surveys and introduce a panel dimension in Addressing public and political concerns re- which a subset of the same households are inter- quires explicitly and transparently communi- viewed over successive years of the survey. Better cating how this set of policies would reduce tracking of the quality of services is also needed, inequality. The perceptions survey fielded by most likely through a combination of survey and the Department of Statistics Malaysia (DOSM) administrative data sources. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 6 01 INEQUALITY CONCEPTS, MEASUREMENT, & WHY IT MATTERS 7 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA T his report examines the measures of inequality and including, for the first trends in different forms of time, a measure of relative income mobility; iii) inequality, in order to un- emphasis on within-group inequality, particularly derstand their drivers and at the intersection of ethnicity and location; iv) identify what Malaysia can a focus on a range of inequality drivers using a do about them. It aims to an- lifecycle approach to unveil how gaps in incomes swer three questions: today are the result of compounding opportunity • How have inequality and mobility evolved over gaps. After the analysis of trends and diagnostics, the last two decades in Malaysia? the report turns to policies. First, it reexamines • What is the state of inequality in Malaysia policy priorities for reducing inequality in the today? context of today’s Malaysia. Second, it outlines a • What are its drivers, and what policies would series of fiscal reforms, including new taxes and help reduce it? better uses of existing spending, that would both reduce the post-pandemic fiscal deficit while The main contribution of the report to the expanding spending on equity- and growth-en- analysis of inequality in Malaysia is its com- hancing policies. Finally, using a bespoke survey prehensiveness, covering the most recent conducted for this report, it looks at Malaysians’ trends, a range of drivers, new perceptions perceptions of inequality and mobility, how big data and policy recommendations. What is they think income gaps are and how concerned new in this report includes: i) updated trends since it makes them, and what policies they think are the Covid-19 pandemic utilizing the most recent most important in addressing inequality. Before household survey data from the 2022 Household delving into those topics, this section outlines Income, Expenditure and Basic Amenities Sur- the motivation, and presents the concepts of in- vey (HIES/BA); ii) these data are included as part equality and mobility, and the framework adopt- of a comprehensive analysis of inequality that ed in the report to understand these concepts. goes beyond the Gini coefficient, using various 1.1 WHY SHOULD WE CARE ABOUT I N E Q UA L I T Y I N M A L AY S I A? F I G U R E 1.1 M A L AY S I A I S O N A PAT H T O A C H I E V E H I G H - I N C O M E S TAT U S , BUT MANY OF ITS PEOPLE ARE NOT 18 100 GNI per capita Thousands of USD 16 90 (Left Axis) 14 80 Sh. High Income people High Income (Right Axis) 70 12 Share of people 60 10 50 8 40 GNI per capita 6 Upper-middle Income 30 4 20 2 Lower-middle 10 0 0 1987 1991 1995 1999 2003 2007 2011 2015 2019 2023f 2027f Source: World Bank staff calculations based on data from World Development Indicators and Household Income, Expenditure and Basic Amenities Survey (HIES/BA). Note: “High-income people” is the share of individuals above the household income equivalent of the HIC threshold (prosperity gap). The prosperity gap projections are based on a distributionally neutral growth assumption. The GNI per capita projections are based on the baseline growth scenario in World Bank (2021a). GNI = gross national income; PG = prosperity gap; USD = US dollars. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 8 M alaysia is on the brink of outcomes, and institutional quality indicators becoming a high-income lag those of aspirational countries (World Bank economy, but the current 2021a). inequality level means that the high-income Sustaining high growth will require stronger level of prosperity is not reform efforts amid stronger global head- yet within reach of many winds Even after achieving high-income status, people in the country. Tackling high inequality Malaysia would have a long way to catch up with is essential for the instrumental reasons of helping more advanced economies—and it would have to eradicate poverty and lift the living standards to do so in a more challenging and uncertain of the country’s poorest, especially in a context of global environment. Malaysia, alongside other slow growth (World Bank 2016a). But while fast-growing economies in East Asia and Pacific, Malaysia’s extreme poverty rate is no longer of faces the major challenges of deglobalization and great concern, the current level of inequality still changing regional trade patterns, a rapidly aging means that the economic prosperity that the population, and climate change (World Bank country wishes to achieve is still a far reality for 2023e). Rapidly evolving technology creates both many. While the country is expected to become a opportunities and challenges for Malaysia. For high-income economy before the end of the instance, digital technologies can promote inclu- decade, its share of high-income people will not sion by enabling existing firms and entrepreneurs rise proportionally. In 2022, close to two-thirds of to serve markets that are currently underserved, Malaysia’s population did not have an income lower costs and encourage innovation and scale level equivalent to the country being a high- economies, but can also result in widening in- income economy. According to the prosperity equality if some workers do not have the right gap (PG) measure, household incomes needed to skills or access to digital infrastructure to com- double on average for everyone in the country to plement existing and future technologies or few reach that status3. By 2028-2030, when Malaysia firms concentrate market power and use digital is expected to have surpassed the high-income technologies to further curtail competition (Re- country (HIC) threshold, less than half of its cord and others 2018). people would be high-income individuals (figure 1.1). Addressing inequality will help Malaysia achieve its aspirations of being an advanced Sustaining high-growth and future prosperi- economy. While there has been much debate ty should not be taken for granted as histori- about the potential trade-off between efficiency cal growth drivers are on a decline. Malaysia’s and equity, cross-country studies on the effect of rapid economic progress was reflected in its inequality on growth are inconclusive (see Ferrei- climb from a low- to middle-income country in ra et al. 2014; World Bank 2016a), and more recent a generation, amid rapid poverty reduction and thinking suggests that there is a complementarity significant improvements in living standards. The between equity-enhancing policies and growth long-term growth path, which was the stron- (World Bank 2005, 2016a).4 The analysis in the gest between 1967 and 1997, was attributed to a World Bank’s 2021 flagship report for the country labor-intensive export-driven model. Large in- (“Aiming High”) outlines six areas where reforms vestment in basic human capital boosted labor are needed for Malaysia to successfully transition productivity and supported the structural trans- to an HIC and sustain equitable growth. These formation of agriculture, while macroeconomic are (1) revitalizing long-term growth, (2) increas- stability was maintained by credible economic ing competitiveness, (3) creating jobs, (4) mod- governance (World Bank 2021a). Growth later ernizing institutions, (5) promoting inclusion, slowed, averaging 4.0 in the decade before the and (6) financing shared prosperity (World Bank Covid-19 pandemic. Average growth rates in re- 2021a). At the macro level, these reforms will help cent years have been lower than the pretransition Malaysia transition to an HIC as they are imple- rates recorded for other countries that managed mented. But they will also affect the income-gen- the transition to high-income status (World Bank erating process and therefore the distribution of 2021a). Moreover, while high for a developing outcomes across households. For instance, a key country, total factor productivity, human capital policy suggested to revitalize long-term growth 3. The PG quantifies the shortfall from prosperity—expressed as the multiple required to achieve a certain prosperity threshold. The PG here uses a Malaysia-specific threshold, which corresponds to the household-equivalent income needed to enter high-in- come status. The higher the PG, the farther away are household incomes from reaching the level for an HIC. 4. World Bank (2016a) uses the term equity-enhancing policies to refer to interventions that reduce inequality. Not all equity-en- hancing policies rely on redistribution in the strict sense of removing resources from one individual or group and giving them to another individual or group. 9 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA is raising the total factor productivity, which will Addressing inequality will help guarantee raise the returns to household assets and, in turn, that Malaysia’s economic transition is inclu- raise household income. Similarly, the returns to sive. Inequality matters for increasing consump- assets also depend on the market structure and tion and lifting people out of poverty. Especially business environment since wages and incomes in countries with large numbers of impoverished from entrepreneurial activity are affected by reg- people or during periods of slow growth, nar- ulations, caps, subsidies, and taxes. In another rowing disparities is essential to ensure the poor example, the inclusion of women in productive benefit from growth and, thus, to maintain the job opportunities, a policy under the third reform pace of poverty reduction (World Bank 2016a). area (“creating jobs”), not only affects the income While Malaysia has a relatively low poverty rate, generated by the current generation of women, addressing inequality also means focusing on the but also affects the returns to the human capital middle part of the distribution—in other words, investments in girls, fostering intergenerational growing the middle class and promoting eco- mobility. nomic security. 1.2 MEASURING INEQUALIT Y T he conceptual basis to un- ferences in assets (e.g., richer households may be derstand inequality across more educated than poorer households, which outcomes, as used in this have less valuable physical or financial invest- report, is an expanded ver- ments), differences in the intensity of use of the sion of the asset framework, assets (e.g., if women stay out of the paid labor which posits that, ultimate- force, they do not use their human capital assets), ly, current inequality is or differences in the returns that the assets gen- determined by differences in who possesses erate (e.g., poor households with unskilled labor assets and how much these assets earn. To receive a lower wage, land with an informal title understand the trends in inequality of outcomes, cannot be used as collateral for a loan, small sav- one has to look at how incomes evolve for differ- ings in cash do not receive interest). Differences ent groups across the distribution and examine in household incomes today drive differences how the different drivers of income have played in consumption (or living standards) as well as for those whose incomes are growing faster or differences in how much households invest in slower. The asset framework posits that house- their children’s health and education (the human holds generate income from their assets—human, capital assets of tomorrow) or how much they physical and financial, and social capital. Each are passing down to them (the financial assets of asset can generate a return in the market: for tomorrow). The extent to which tomorrow’s in- example, the labor and knowledge from human comes depend on this accumulation of assets to- capital earn a wage, and physical and financial as- day, is a major driver of how much economic mo- sets earn an income (rent from land and housing, bility individuals experience in their lifetimes or interest or dividends on investments) (arrow 1 in their children experience relative to their parents. figure 1.2). Current inequality can arise due to dif- F I G U R E 1. 2 I N C O M E G E N E R AT I O N Shocks Taxes and transfers ASSET FRAMEWORK Households spend T O U N D E R S TA N D 1 RM 2 their income on consumption INEQUALIT Y (determining Assets Income Consumption inequality today) Households have Returns to assets different quantity and determine how much quality of assets income is generated Unspent income can ▪ Human capital through assets be saved or invested ▪ Physical and ▪ Labor income RM in assets for the Financial capital ▪ Interests and rents concurrent and next ▪ Social capital generations Investment RM 3 Source: Based on Bus- solo and López-Calva Inter/ intra generational transmission of income (2014) and World Bank generation through accumulation of assets (mobility) (2016b). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 10 Poorer households are not only less able to invest in income-generating assets for the future, but also less able to save for future shocks. The report uses income as a proxy for wel- equality changes over time as it affects how much fare and to monitor the trends and drivers of households are able to save and invest in further inequality. Outcomes can be measured in the accumulation of assets. income or consumption spheres. Income is the indicator used in Malaysia and many upper-mid- Not all inequality is the same. First, there is dle and high-income countries to monitor wel- a distinction between market inequality and fare, and, consequently, the report follows this post-fiscal inequality (after taxes and trans- approach.5 Income is also the most important fers). The distinction between market inequality determinant of consumption, and using income and post-fiscal inequality allows highlighting the appeals since it allows studying the drivers of in- role of taxes and transfers, individually and joint- equality. Households spend their income on ei- ly, and thus helps to consider the potential policy ther consumption or on saving for future invest- changes that can be introduced to promote equi- ments (arrow 2 in figure 1.2). Those with higher ty. For instance, Belgium, Spain, Japan, the United income can save and invest more, meaning higher States, Türkiye, and Chile have a similar degree of quantity or quality of the assets they will have in inequality before taxes and transfers, but a very the future (or for the next generation) (arrow 3 different one after (Hasell 2023). Market income in figure 1.2), which, in turn, will generate higher inequality tends to be higher, and this is not nec- returns, thus reinforcing the cycle of inequali- essarily problematic insofar as it reflects different ty. In contrast, poorer households spend much levels of effort, talent, entrepreneurship, and even more on basic consumption and current living luck. And while there is no set benchmark, the expenses. They are not only less able to invest in degree to which countries reduce market income income-generating assets for the future, but also inequality through the fiscal system varies and less able to save for future shocks. The cycle of tends to be greater in high-income economies inequality is also affected by households’ demo- compared with low- and middle-income ones graphic composition. It is common for richer (World Bank 2022c). households to be smaller and have fewer depen- dents (especially children); this means that in- The income-generating process is affected comes need to finance the consumption of fewer by shocks and by policies used for redistri- household members, which in turn leaves a larger bution or to protect households from such share to save or invest. But as countries develop, shocks. Shocks can reduce household income at fertility rates fall and life expectancy increases. all points of the income-generating process de- This demographic transition can affect how in- scribed above, and the incidence of shocks may 5. See appendix A for details on the measurement of welfare in Malaysia and the differences between the official Department of Statistics Malaysia (DOSM) estimate and the World Bank methodology used throughout this report. 11 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA be uneven across the distribution. For example, For example, investments in better health and in many Latin American cities, poorer households education for disadvantaged children support living in informal settlements are more prone to human capital accumulation and raise their in- landslides and loss of physical and human capi- come-earning ability and can thus lower inequal- tal when they occur. Fiscal policy can lower or ity in the future. It is important to note, howev- heighten inequality through the allocation of er, that the quality of such investments, which the taxes that households pay and the transfers is often skewed, influences how much different that they receive. For example, social assistance households actually benefit from public spend- transfers lower inequality if they are targeted to ing. Investments in education, for instance, will the poor and the benefits distributed are ade- not lower inequality if poorer children receive quate. On the other hand, some fiscal policies can lower-quality education than richer children and, heighten inequality if they mostly benefit richer consequently, are less productive and earn lower people. For instance, unemployment insurance returns on their skills. can provide some temporary relief for house- holds facing a job loss, but informal workers, who The report also considers people’s views tend to be poorer, are often unlikely to qualify for it. about the distribution of outcomes. Percep- tions of inequality may or may not match ob- There is also a distinction between inequality jective measures, but they still matter because of outcomes and inequality of opportunities. people’s perceptions form the basis for their at- Using a lifecycle approach, the report tries to un- titudes, behaviors, and support for policies. Per- veil how gaps in incomes today are a result of op- ceptions of high inequality are linked to concepts portunity gaps. Income inequality today (inequal- of fairness. The same level of income inequality ity of outcomes) has roots in the opportunities can be deemed acceptable or not depending on that people have to accumulate productive as- whether people perceive it to be caused by effort sets. These opportunities start to be shaped early, and skills, or by low opportunities for mobility or and, because of this, many are a result of circum- undue market power. stances beyond an individual’s control (such as their ethnicity, region or district of birth, or par- Other dimensions of inequality are outside ents’ income and education). Even before birth, the scope of this report; an important reason an adverse prenatal environment can lead to for this is the difficulty of obtaining quality worse birth outcomes and early childhood health data that can shed light on these dimensions. (Almond and Currie 2011). Early childhood health An important area where only limited analysis and nutrition, in turn, affect children’s learning is possible, due to difficulty of obtaining quali- once they start school, and these also reflect in ty data, is the distribution and concentration of final educational attainment. Thus, labor market wealth. Wealth is as a measure of long-term wel- outcomes and the returns earned by people are fare: it can be passed on to future generations and affected by a chain of opportunities that com- used to cushion shocks, and can earn a return pound over the lifecycle (figure 1.3). A progres- in the market; but it can also be an indicator of sive fiscal policy can have immediate impacts on political influence, power, and market concentra- inequality, but it can also affect inequality in the tion, among others. The report also does not look future by helping to finance public investments at financial and land assets, or social capital, and and services to give poorer children a fair start. returns to these types of capital. F I G U R E 1. 3 O P P O R T U N I T I E S S TA R T E A R LY I N L I F E A N D COMPOUND OVER TIME � Prenatal Birth Early Basic Learning Educational Labor Income childhood literacy outcomes attainment market health A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 12 1.3 MOBILIT Y M obility is about the dynam- in life, with the accumulation and ownership of ics of the income genera- productive assets (physical, human, financial, and tion process—the feedback social). But when income inequality is high, those loop in the asset frame- at the bottom of the distribution cannot save work. While the report fo- and invest, limiting the accumulation of assets. cuses on inequality of out- This results in low mobility for the concurrent comes, understanding those generation (intergenerational mobility) and low outcomes requires analyzing the distribution of mobility for the next generation as the opportu- opportunities and the mobility of people in the nities for the children of these poorer households outcome distribution. Inequality and mobility are are constrained (intragenerational mobility). This related but distinct concepts. The recent World negative relationship between income inequality Development Report stresses the importance of and mobility is seen across countries (figure 1.4). tracking mobility; countries with greater social This empirical evidence shows that the greater mobility are better at developing skills and using the inequality there is in a country, the greater the talent, both key aspects for middle-income coun- difficulty in progressing from one position to an- tries to infuse, innovate, and grow (World Bank other over the course of one generation (Narayan 2024b). Inequality of outcomes has roots early et al. 2018). F I G U R E 1 .4 ZA 0.6 C O R R E L AT I O N GT BET WEEN INCOME 0.5 BO PA BR MX CLRW INEQUALIT Y AND EC GN INEQUALIT Y OF 0.4 US AR GH OPPORTUNIT Y Inequality AU GBLV PT RO ES BGIT LT FRGR PL KR HRCH IE EE 0.3 DE ATCY MT HU LU DK SE FI NL SK CZ BE IS NO SI 0.2 0.1 Source: https://www. equalchances.org . World Database on Equality of 0 Opportunity and Social 0 0.1 0.2 0.3 0.4 0.5 0.6 Mobility Inequality of opportunity Low mobility is harmful for economic growth unchangeable, decided by, for example, gender, and is intrinsically unfair as it reflects limited income status, or racial background, the resulting opportunities for some people because of cir- feeling of powerlessness and lack of resources cumstances beyond their individual control. for improvement can hinder goal setting and in- Some level of inequality is part of the incentive vestment in the future. Low mobility, as it reflects structure in an economy (to study, work hard, in- limited opportunities for some people, can be vest, innovate, and take risks), or simply reflects thought of as intrinsically unfair. Outcome differ- the different levels of talent and effort among ences caused by circumstances beyond an indi- individuals, but it is increasingly recognized that vidual’s control (e.g., ethnicity, place of birth, or low mobility is detrimental for growth. To begin parents’ education) are unfair and can be seen as with, in the absence of mobility, incentives do a waste of human and economic resources in an not work as well to change behavior. For exam- economy: “If many potential scientists, engineers, ple, young people may decide not to continue and artists cannot access a decent education or their education if there are no opportunities to basic healthcare early on because of the circum- generate more income through better education. stances of their birth, that ultimately creates a Similarly, low mobility can affect people’s aspira- cost for everyone in terms of wasted human and tions and thus behaviors; when individuals per- economic potential” (Ferreira, quoted in World ceive their places in the social order as fixed and Bank 2019). 13 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Mobility is important to understand percep- over time. Mobility can be measured in mone- tions of inequality. Low mobility in a context of tary terms (economic mobility) but also in non- high inequality can be detrimental for growth as monetary terms (educational or class). In general, well as for social cohesion and stability. A high- panel surveys that follow the same households er degree of income inequality may be more ac- over time are needed to make statements about cepted if the degree of mobility is high: the rich economic mobility. But these data are scarce, have higher incomes than the poor, but someone even in upper-middle-income countries. In the who is poor now (maybe by birth) can work hard absence of panel data, synthetic panels7 or retro- and become rich, while someone who is rich spective information can help capture mobility. now (maybe by birth) is not guaranteed to stay Absolute mobility refers to where a generation is at the top of the distribution no matter what. But placed relative to the previous generation with a high inequality–low mobility society seems less respect to a fixed benchmark. In rapidly growing fair; not only are incomes at the top much higher countries like Malaysia, absolute mobility tends than incomes at the bottom, but people who are to be high because as poverty and vulnerability at the bottom are stuck there. This combination fall over time, most people become better off affects incentives to invest in human capital and than their parents. Relative mobility is about entrepreneurial activities, for example, and the changes relative to other households across the sense of unfairness from layering low mobility on income distribution. Thus, even if everyone was top of high inequality contributes to a break in richer or more educated than their parents, rela- social cohesion and public unrest, which create tive mobility can be low if the initially rich move instability and harm growth (e.g., during the Arab away faster than the poor and the ranking in the Spring).6 distribution is preserved over time. In terms of equality of opportunity, relative mobility is the Mobility can be absolute or relative, and the key. Relative mobility is important as it more measurement of mobility requires data that closely reflects inequality of opportunities and can help track individuals (and their children) affects perceptions of inequality. 6. High inequality has been correlated with lower social cohesion and higher political instability (Gupta 1990; Keefer and Knack 2002). For the period prior to the Arab Spring, the people in these countries reported a growing sense of despondency, meaning people increasingly believed their opportunities for success were shrinking, and success was less correlated with effort, and there was a sense that prospects for future generations were worsening (Arampatzi et al. 2015; Chattopadhyay and Graham 2015). 7. See Dang et al. 2014. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 14 02 INEQUALITY IN MALAYSIA OVER THE PAST TWO DECADES 15 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA O ver the past two decades, nificant progress, Malaysia’s Gini remains higher income inequality in Ma- than that seen for the advanced economies that laysia has been falling, but it wishes to emulate. the decline has stalled in recent years. Inequality as Current inequality in Malaysia does not measured by the Gini index stand out in the region, but it is higher com- declined from a very high pared with that in more advanced econo- level of above 50 in the 1970s and 1980s to just mies. The current level of inequality in Malaysia, under 40 today (figure 2.1). The sharpest decline as measured by the Gini index, is higher (Gini was in the 1980s, when the Gini fell by an average of 39) compared with both recently transi- of 0.6 points per year. While inequality has kept tioned high-income countries (HICs) (mean of falling, the decline has stalled more recently. In 31)9 and established HICs (mean of 30)10 (figure the roughly past two decades on which this re- 2.2). Among regional peers, Malaysia has the port focuses (2004–22),8 inequality as measured third-highest Gini, after the Philippines and Thai- by the Gini index fell from 45 (in 2004) to 40 (by land, although other countries in the region use 2014) and remained at that level through to 2019, consumption to measure welfare, which can un- before the COVID-19 pandemic, and further de- derestimate the level of inequality relative to the clined slightly, to 39, according to the most recent countries that use income.11 data from 2022 (figure 2.1). While this is not insig- F I G U R E 2 .1 M A L AY S I A G I N I I N D E X L O N G - R U N T R E N D , 1 9 7 0 – 2 0 2 2 60 55 50 Gini index 45 40 35 30 1970 1977 1984 1991 1998 2005 2012 2019 2022 Department of Statistics Malaysia (gross income per household) World Bank (net income per capita) Source: World Bank staff calculations based on official estimates from DOSM; data from the World Bank Poverty and Inequality Platform (https://pip.worldbank.org/); and calculations based on data from the HIES/BA. Note: DOSM calculates the income Gini using the total (gross) household income; the World Bank uses per capita (disposable) household income. The World Bank series estimate is comparable for the period 1984–97 and 2004–22. 8. The analysis of trends and drivers of inequality in this report focuses on roughly the last two decades since this is the period for which microdata from the HIES/BA are available. 9. Aspirational comparator countries that reached high-income status in the past 30 years are Chile, the Czech Republic, Estonia, Hungary, the Republic of Korea, Lithuania, Latvia, Poland, Portugal, Slovak Republic, Slovenia, Argentina, Croatia, Oman, Pana- ma, Puerto Rico, Saudi Arabia, Trinidad and Tobago, Uruguay, Mexico, and Türkiye. 10. Organisation for Economic Co-operation and Development (OECD) countries. 11. Most Asian countries use consumption to measure welfare, and consumption-based Ginis tend to be lower than income-based Ginis because income includes savings and richer people have a higher marginal propensity to save. Inequality in Thailand is much higher when using income instead of consumption. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 16 FIGURE 2.2 I N E Q U A L I T Y I S H I G H E R I N M A L AY S I A T H A N I N HIGH-INCOME COUNTRIES 60 50 40 Gini index 30 20 10 0 MMR THA* IDN LAO MYS PHL THA SVK SVN POL SWE AUT VNM PHL* CZE HRV HUN POL EST DNK IRL DEU CHL KOR PRT LVA LTU URY ARG PAN BEL NLD NOR FRA USA TUR MEX ASEAN Economy that recently OECD achieved high-income status Source: World Bank staff based on data from the World Bank Poverty and Inequality Platform (https://pip.worldbank.org/) and Department of Statistics Malaysia (DOSM) Household Income, Expenditure and Basic Amenities Survey (HIES/BA). Note: Countries in dark use income for the welfare aggregate; countries in light shade use consumption. Malaysia and the Philippines use income for the official welfare aggregate, but they also have consumption measures available (marked with *). Malaysia’s Gini is calculated from the 2022 HIES and based on per capita income rather than the official estimate from DOSM using household total income. ASEAN = Association of Southeast Asian Nations; OECD = Organisation for Economic Co-operation and Development. Trends in inequality are similar when consid- (figure 2.3), whereas the T10/B40 ratio only con- ering different measures. The level of inequal- siders the top 10 percent and bottom 40 percent ity is about three points higher for Malaysia, but of the distribution. Several other commonly used the trend is similar when looking at the Gini for measures of inequality consistently show the market income, that is, when excluding taxes and same pattern: a decline from 2004 to 2014, with transfers. Different measures place emphasis on inequality remaining roughly unchanged since different parts of the distribution; for example, then (figure 2.4). the Gini places more emphasis in the middle part FIGURE 2.3 60 M A L AY S I A G I N I I N D E X , 2004–22 55 50 Gini index 45 40 Source: World Bank staff calculations 35 based on DOSM HIES/BA. Note: Income is defined as per capita 30 household income. Market income 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 Gini is calculated using per capita household income before taxes and transfers, while the net income Gini Gini (net) Gini (market) includes them. 17 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E 2 .4 60 5.0 INEQUALIT Y TREND WITH 55 4.5 DIFFERENT MEASURES, Gini, GE, Atkinson 2004–22 50 4.0 45 3.5 40 3.0 Ratio 35 2.5 Source: World Bank staff calculations 30 2.0 based on DOSM HIES/BA. 25 1.5 Note: The Theil (GE0 or mean log deviation) and the interdecile ratios have no upper bound, whereas the 20 1.0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Atkinson and the Gini range from 0 to 100. The higher the value in the sensitivity parameter in the Atkinson, Gini (net) Theil GE(0) Atkinson(1) the more sensitive the index is to the bottom of the income distribution. Atkinson(2) p90p40 ratio (RHS) The stalling of inequality reduction has co- rich, the sentiment is more prevalent among incided with people’s growing perception of those who felt they were in the poorest group. a widening gap between the rich and poor. The perceptions survey reveals that Malaysians In the survey of public perceptions of inequality discussed the difference between the poor and conducted in early 2023, 70 percent of the re- the rich most often, more commonly than oth- spondents12 perceived that the gap between the er dimensions (e.g., spatial differences or socio- rich and the poor today was “wide” or “very wide.” demographic differences). When the view on When asked about the same gap a decade ago (in the widening income gap is assessed by income 2013), that percentage fell to half of the respon- distribution according to the respondents’ per- dents (figure 2.5). A similar pattern is observed re- ceived income class, nearly half of the individuals gardless of respondents’ ethnicity and where they in the poorest group believe that the income gap are placed in the income distribution. today was very wide. This is three times higher than the share of people with the same view who While most Malaysians are concerned about felt they belonged to the richest group (figure 2.6). the widening gap between the poor and the 12. Survey respondents are the heads of households. See appendix B for details on the Survey of Malaysians on Perceptions of Inequality and Intergenerational Mobility 2023. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 18 F I G U R E 2 .5 PERCEPTIONS OF INCOME GAP 10 YEARS AGO AMONG ACTUAL INCOME CL ASSES Ten years ago (2013) 13% 11% 9% 10% 11% 39% 38% 38% 38% 38% 33% 34% 35% 37% 35% 15% 15% 16% 14% 14% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 (Q1) (Q2) (Q3) (Q4) (Q5) Very wide Wide Moderate Small Very small F I G U R E 2 .6 P E R C E P T I O N S O F I N C O M E G A P T O D AY A M O N G A C T U A L AND PERCEIVED INCOME CL ASSES Today (2023), by actual income class Today (2023), by perceived income class 25% 17% 22% 27% 27% 30% 29% 30% 35% 33% 35% 44% 47% 46% 45% 44% 45% 47% 45% 47% 46% 30% 26% 27% 24% 29% 24% 21% 18% 14% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Very wide Wide Moderate Small Very small Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: What is your view on the current income gap between the rich and the poor? What is your view on the income gap between the rich and the poor in the past 10 years (wide/very wide/moderate/small/very small)? Actual quintiles are derived from per capita household income recorded in the HIES/BA (2022) and matched to the Survey of Perceptions of Inequality. Perceived quintiles are based on the responses to the following question: “Which income class his/her household belongs to (if all households in Malaysia were divided into 10 equal groups, with 1 being the poorest group and 10 being the richest group)?” People’s perceptions of their welfare and of Malaysians felt richer than they were based on an the trend in inequality may not match reali- income metric, while richer Malaysians felt that ty. In Malaysia, while the relative Gini was falling, they were less well off than they actually were the mean absolute gap (absolute Gini) was rising (figure 2.8). Notably, across all income groups, (figure 2.7). What people perceive with regard to the majority of people felt that they were in the the evolution of inequality may be more close- middle of the income distribution. The extent ly related to absolute rather than relative gaps. to which these (mis)perceptions affect their at- There can also be a mismatch between subjec- titude toward inequality is important for public tive and objective measures of welfare. The per- support toward policy reforms. ceptions survey, for instance, shows that poorer 19 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 800 60 FIGURE 2. 7 ABSOLUTE VERSUS 700 55 R E L AT I V E G I N I , 2 0 0 4 – 2 2 600 Relative Gini index 50 Absolute Gini 500 400 45 300 40 Source: World Bank staff calculations based on the DOSM HIES/BA. 200 Note: Absolute Gini (S-Gini) is half of 35 the mean income difference. 100 0 30 2004 2007 2009 2012 2014 2016 2019 2022 Absolute Gini Relative Gini FIGURE 2.8 M AT R I X O F A C T U A L A N D Actual income class (per capita income) PERCEIVED INCOME CL ASSES OF HOUSEHOLDS Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 12% 6% 4% 2% 2% Perceived income class Source: DOSM-World Bank Survey of Perceptions of Inequality and Quintile 2 38% 34% 30% 27% 22% Intergenerational Mobility (2023). Notes: If we divide all households in Quintile 3 44% 50% 52% 53% 50% Malaysia into 10 equal groups, with 1 being the poorest group and 10 being the richest group, which income class Quintile 4 6% 10% 13% 17% 24% does your household belong to? (1 is the lowest or poorest income class, while 10 is the highest or richest Quintile 5 0% 0% 0% 1% 2% income class.) Inequality would be even higher if the in- lid and Yang (2021) documented a 7-percent- comes of the richest Malaysians were ac- age-point gap between the estimates of the top 10 counted for. Estimates derived from household percent share of income derived from household surveys are limited by the coverage of such data surveys and that corrected for underreporting in in many countries, in particular because some 2014. This gap appears to be narrowing, from a populations and sources of income are especial- peak of 11 percentage points in 2007, although it ly poorly captured in surveys.13 Top incomes, or is difficult to make a conclusive assertion regard- the incomes of the richest households, are one of ing whether this was due to a change in report- those areas. In 2019, and according to survey data, ing or a true decline in the gap between the two the top 10 percent of households held 30 percent sources.14 Nonetheless, the existence of this gap of the income. But this is an underestimation of means that if the missing top incomes were ac- the actual concentration of income at the top counted for, then inequality in Malaysia would be of the distribution. A study for Malaysia by Kha- even higher than survey data alone show. Using 13. See, for instance, Yonzan et al. (2022); Alvaredo and Londono Velez (2013); Atkinson et al. (2010), and Jenkins et al. (2011), among many others. 14. Interpreting changes over time in top income shares or adjusted inequality estimates is problematic. Changes in levels (trends over time) may be due to changes in underlying inequality or top income shares, but may also be due to changes in the adminis- trative or other data used to estimate these measures. For example, US income tax data have been used to show that top income shares have increased significantly in the past 50 years (Piketty and Saez 2003; Piketty, Saez and Zucman 2018), but recent work suggests that these estimates can be biased due to changes in the tax base, marriage and divorce, and missing income sources (Auten and Splinter 2024). Correcting for these biases suggests that top income shares are lower and have increased less since 1980 than in the earlier studies. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 20 individual administrative tax data for 2016, 2019, mate that the Gini for individuals with employ- and 2022 to incorporate data on high-earning ment income would be 1–2 points higher in those households missing from the survey, and the ad- years than what the survey data alone show.16 This justment proposed by Alvaredo (2011),15 we esti- is a much smaller adjustment than seen in many F I G U R E 2 .9 S H A R E O F T O TA L W E A LT H H E L D B Y T H E R I C H E S T 1 0 PERCENT OF HOUSEHOLDS (PERCENT ) 100 90 80 70 Percentage of total wealth 60 50 40 30 20 10 0 Russian Fed. Turkiye Vietnam Korea, Rep. Poland Taiwan, China Hong Kong SAR, China Indonesia Philippines Czech Republic China Germany Singapore Thailand United States Egypt, Arab Rep. Brazil Peru Switzerland Argentina Malaysia South Africa Sweden Denmark Chile Saudi Arabia Norway Colombia Mexico Austria United Arab Emirates Ireland Portugal Canada New Zealand Greece Spain Netherlands Finland United Kingdom France Italy Australia Japan Belgium Source: Credit Suisse 2014. other countries—Alvaredo estimated an 11 point ily evaded. Although the labor income share of adjustment for the United States and a 6–8 point the gross domestic product (GDP) in Malaysia adjustment for Argentina. The small adjustment has been growing over the past two decades, it for Malaysia likely reflects that the tax microdata remains lower than regional, aspirational, and do not capture much capital income among the Organisation for Economic Co-operation and rich households (averaging just 12 percent of the Development (OECD) comparators (figure 2.10). total income in the data, compared with about 60 This means the capital income share (the inverse percent in national estimates; see figure 2.10). of the labor income share) is high. Moreover, cap- ital ownership is highly concentrated. While data Further, wealth inequality is highly concen- are scarce (and what there is has limitations), the trated. Wealth is the result of cumulative invest- available evidence indicates that wealth inequali- ment and savings decisions over time. Accumu- ty is even more highly concentrated than incomes lated wealth generates even higher incomes in in Malaysia, as in other Southeast Asian countries the future, driving inequality higher. For example, (figure 2.9). The richest 10 percent of Malaysian the returns to financial assets have been very households held 70 percent of wealth (Credit high in Malaysia, but these assets are owned only Suisse 2014), meaning that this outsized income by the wealthy, and taxes on capital income are share benefits the richer few. Further, data from often lower than taxes on labor17 and more eas- the Forbes magazine18 shows that the aggregate 15. Whereby the Gini coefficient, G can be approximated by G = (β−1)/(β+1) * PS + G*(1−P)(1−S) + S−P, where P is the share of the top income population with a share S of the total income, taken from tax data or some other non-survey data, and G* is the Gini coefficient for the rest of the (1-P) population from the survey. 16. The adjustment is relatively stable regardless of whether the tax and survey data are truncated; hence, P = the top 1, 2, 3, or 4 percent of the income distribution. 17. Malaysia does not have capital gains tax except property gains tax. 18. These data have limitations, for example, potential underreporting of wealth, or the fact that the wealth of billionaires moves closely with the stock market. 21 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E 2 .1 0 L A B O R I N C O M E S H A R E O F G D P, 2 0 2 0 70 Labor share of income (percent) 60 50 40 30 20 10 0 Philippines Viet Nam Argentina Slovakia Lithuania Korea, Rep. Latvia Lao PDR Malaysia Myanmar Thailand Panama Hungary Poland Czechia Portugal Estonia Croatia Indonesia Ireland Mexico Türkiye Norway Denmark Japan USA France Germany Belgium Chile ASEAN OECD Transitional Source: ILOEST database; https://ilostat.ilo.org/methods/concepts-and- definitions/ilo-modelled-estimates/. Note: ASEAN = Association of Southeast Asian Nations; OECD = Organisation for Economic Co-operation and Development. F I G U R E 2 .1 1 20 A G G R E G AT E N E T 18 W O R T H O F M A L AY S I A’ S Percent of GDP 16 BILLIONAIRES AS A 14 S H A R E O F G D P, 2 0 0 1 – 2 3 12 10 8 Source: World Bank staff 6 calculations based on Forbes (2023b). 4 Note: GDP = gross domestic 2 product. 0 2011 2001 2003 2005 2007 2009 2013 2015 2017 2019 2021 2023 net worth of Malaysia’s billionaires as a share of licenses, a reputation of illegality, or potential GDP has steadily increased in the past two de- monopolistic practices. In contrast, non-rent- cades, reaching 13 percent of GDP in 2023,19 thick sectors have more limited interaction with and is high relative to that in the Association of the state, the regulator has a good reputation, and Southeast Asian Nations (ASEAN), OECD, and there is little anecdotal evidence of illegal practic- transitional countries (figure 2.11 and 2.12). es. Of Malaysia’s 50 richest individuals in 2024, 28 made their wealth from sectors considered rent And wealth is concentrated in sectors con- thick (real estate, construction, infrastructure or sidered “rent thick.” Gandhi and Walton (2012) ports sector, media, cement, and mining). Simi- propose a classification of economic activity into larly, wealth from rent-thick sectors accounts for “rent-thick” sectors, which are sectors where 72 percent of the aggregate net wealth of these there is a pervasive role of the state in providing individuals.20 19. The aggregate net worth of Malaysia’s 50 richest individuals accounted for 19 percent of GDP, according to the Forbes 50 Richest Malaysians data (Forbes 2023a), which were not available over time to track trends. 20. Note that when an individual has multiple industries as their wealth source, it is not possible to distinguish from the data how much of their net worth comes from a rent-thick sector, or whether they have any income from a rent-thick sector; this is because the Forbes database does not disaggregate net worth by industry. Because the Gandhi and Walton (2012) classification is done for India (drawing from a KPMG survey on corruption perceptions in this country), we conduct a sensitivity analysis adding some sectors that could be considered “rent thick” in Malaysia. Industries where there is a substantial presence of government-linked companies (GLCs) significantly contribute to the rent-thick nature of industries in Malaysia. Given the strong presence of GLCs, we add the following sectors to the rent-thick list for a sensitivity analysis: natural resources (logging, palm oil), oil/energy, finance and banking, automotive industry, and ship building. Using this alternative definition slightly raises the share of wealth of the total net worth of Malaysia’s 50 richest individuals from rent-thick sectors, from 72 percent to 80 percent, mainly due to the addition of four of the 50 richest individuals who gain their wealth from rent-thick sectors. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 22 F I G U R E 2 .1 2 B I L L I O N A I R E S ’ W E A LT H A S A S H A R E O F G D P, 2 0 0 1, 2 0 1 0, AND 2022 Share of billionaire wealth to GDP, Share of billionaire wealth to GDP, Share of billionaire wealth to GDP, selected countries (2001) selected countries (2010) selected countries (2022) 15 20 30 15 Percent Percent Percent 10 20 10 5 10 5 0 0 0 THA PHL MYS BEL DNK NOR CHL IRL MEX FRA DEU TUR USA SWE KOR ARG PRT IDN THA PHL MYS BEL DNK NLD NOR AUT FRA TUR DEU IRL MEX USA CHL SWE ARG POL KOR PRT CZE VNM IDN PHL MYS THA BEL NLD TUR NOR IRL MEX CHL AUT DNK DEU USA FRA SWE HUN PRT ARG SVK POL EST ERY KOR CZE ASEAN OECD Transitional ASEAN OECD Transitional ASEAN OECD Transitional Source: World Bank staff calculations based on Forbes (2023b). Note: ASEAN = Association of Southeast Asian Nations; OECD = Organisation for Economic Co-operation and Development. High inequality in Malaysia is reflected in the needed to increase 4.6 times for everyone to fact that while the country is nearly a high-in- reach an HIC income level in 2004, and this fell come economy, over 70 percent of its people to 2 times by 2022 (figure 2.13). This trend in the still live far from that level. Kraay et al. (2023) PG roughly follows the inequality trend. Fur- propose a new index, referred to as the prosper- ther, although average growth is very important ity gap (PG), which quantifies the shortfall from to improvement in the PG, inequality also mat- prosperity—expressed as the multiple required ters. This can be seen when decomposing im- to achieve a certain prosperity threshold. The provement in the PG into improvement due to PG here is calculated following the method- changes in mean incomes and improvement due ology in Kraay et al. (2023) but using a Malay- to changes in inequality22 (figure 2.14). Reduction sia-specific threshold, which corresponds to the in inequality contributed 1.6 percentage points to household-equivalent income needed to enter the reduction in the PG between 2004 and 2014 high-income status.21 The farther away house- but only 0.1 points between 2014 and 2019. Put- hold incomes are from reaching the level for an ting it differently, for the same mean growth, the HIC, the larger the PG. While virtually the entire PG could have declined even further, by another Malaysian population was below the threshold in 30 percent, had inequality continued declining 2004, the share had fallen to 71 percent in 2022. between 2014 and 2019. In the three years since 2019, which spanned the COVID-19 pandem- Malaysia has made substantial progress in ic, when growth was subdued, the contribution raising people’s incomes to an HIC standard, of inequality reduction was again visible, but its but this could have been done much faster contribution was still only half as much as that of had inequality continued falling. By the PG mean growth. measure, average Malaysian household incomes 21. Calculated as the threshold to enter high-income status using the market exchange rate and adjusting for the difference between the average income from a household survey and GNI per capita based on national accounts. This is equivalent to US$13,205 in 2021 (roughly RM 2,000 per month per person in 2017 prices). This differs from the global version of the PG, which uses a prosperity standard of US$25 per person per day in 2017 Purchasing Power Parity (PPP) terms. 22. Growth (improvement) of prosperity gap: - LN(W) = - LN(I) + LN(mean). 23 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E 2 .1 3 P R O S P E R I T Y G A P I N M A L AY S I A ( H I G H - I N C O M E S TAT U S T H R E S H O L D I N 2 0 1 7 $ P P P ), 2 0 0 4 – 2 2 5.0 4.6 4.5 4 4.0 3.8 Prosperity gap 3.5 3.1 3.0 2.5 2.5 2.2 2 2.0 1.8 1.5 1.0 0.5 0.0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 Source: World Bank staff calculations based on the DOSM HIES/BA. F I G U R E 2 .1 4 D E C O M P O S I T I O N O F T H E P R O S P E R I T Y G A P I N M A L AY S I A , 2004–14, 2014–19, 2019 –2 2 6 1.6 0.1 Annual change (perecent) 4 6.1 4.9 2 0.9 2.7 4.5 4.8 1.8 0 2004–14 2014–19 2019–22 Contribution of change in inquality (%) Contribution of mean growth Source: World Bank staff calculations based on the DOSM HIES/BA. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 24 2.1 EXPLAINING INEQUALIT Y TRENDS WITHIN THE COUNTRY I nequality declined from higher for the 60th to 80th percentiles of the 2004 to 2014 when growth distribution. The GIC from 2019 to 2022 shows was pro-poor, but from slightly pro-poor income growth, although less 2014 on, income growth than in the early period, but overall growth slow- was not only lower for down is significant since the pandemic (average most households, it was household per capita income growth was closer also much more similar. to 2 percent, significantly below that in 2002–14). The trends in aggregate inequality indices de- scribed in the previous section reflect the pattern Despite pro-poor growth, income is still of household income growth across the income highly concentrated at the top. Incomes for distribution and for various subgroups of peo- the poor and middle parts of the distribution ple within Malaysia. Growth incidence curves have grown faster, but because they started from (GICs) show how the rate of growth varies along a low base, the absolute gaps remain. This can be the percentiles of the income distribution and al- seen in the share of income held by each quin- low seeing whose incomes have grown more or tile of the distribution (figure 2.16). The bottom less within a period.23 From 2004 to 2014, when 20 percent of people in Malaysia hold less than inequality was declining, the GIC shows high- 6 percent of income today. This has risen slightly er-than-average growth rates at the bottom of from 4.6 percent in 2004, but it is still a low figure. the distribution (figure 2.15). From 2014 to 2019, In contrast, the share of income held by the top before COVID-19, income growth was not only 20 percent of the distribution fell from 46 per- lower for most households, it was also much cent in 2004 to 41 percent in 2022.24 more neutral; if anything, growth was slightly F I G U R E 2 .1 5 Annual economic growth rate (percent) 8 GROW TH INCIDENCE 6 CURVES, 2004– 4 14, 2014–19, 2019–22 2 Source: World Bank 0 staff calculations based on data from the World 1 20 40 60 80 100 Bank Poverty and Income percentile Inequality Platform (https://pip.worldbank. 2004–14 2014–19 2019–22 org/). DOSM HIES/BA. F I G U R E 2 .1 6 100 1.3 INCOME SHARES 90 1.3 ACROSS THE Share of total income (percent) 80 1.2 DISTRIBUTION, Growth index (2004=1) 70 1.2 2004–22 60 1.1 50 1.1 40 1.0 30 1.0 20 0.9 Source: : World Bank 10 0.9 staff calculations based on data from 0 0.8 the World Bank 2004 2007 2009 2012 2014 2016 2019 2022 Poverty and Inequality Platform (https://pip. Income quintile worldbank.org/). Q1 Bottom Q2 Q3 Q4 Q5 Top 23. A downward-sloping GIC means pro-poor growth. 24. These figures do not account for missing top incomes. 25 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Labor income and household size were the so did incomes from wages and self-employment main contributors to the decline of inequal- (figure 2.17). Transfers, which include private and ity in the first decade but ceased to be after public transfers, increased inequality in this de- 2014. A Shapley decomposition25 shows the cade.26 Between 2014 and 2019 and 2019 to 2022, contribution of different sources to the change changes in inequality were very small; hence, all in inequality. Household demographics played a components had much more muted contribu- significant role in the decline of inequality in the tions. Nonetheless, while household size con- first decade, especially because household sizes tinued to be a positive force for inequality, labor declined faster for poorer households, making incomes changed and became an inequality in- them better in per capita terms (see box 2.1), and creasing factor. F I G U R E 2 .1 7 2004-14 2014-19 2019-22 SOURCES OF 1.5 INEQUALIT Y Sources of inequality (Gini points) 1.0 CHANGE, GINI POINTS, 2004– 0.5 14, 2014–19, 0.0 2019–22 -0.5 -1.0 -1.5 -2.0 Household size Share adult Share employed Wages Self employment Property Transfer Source: World Bank staff calculations based on DOSM HIES/BA. F I G U R E 2 .1 8 50 L ABOR INCOME 45 SHARE FOR M A L AY S I A , 2 0 0 4 - 2 2 40 35 30 Source: : International Per cent Labor Organization (ILO), 25 Penn World Table (PWT), and DOSM. 20 Note: Sources differ 15 primarily on the treatment 10 of self-employed incomes. ILO and PWT use the share 5 of employees’ compensa- tion plus the labor income 0 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 of the self-employed. DOSM excludes self-em- ployed incomes. ILO PWT DOSM The share of labor income in GDP also stag- ity flattened out), and has declined sharply over nated since 2014. Sources differ on the levels COVID-19 due to lower incomes in the service but show a similar trend (figure 2.18): the labor sectors. These trends explain why labor income share was growing faster between 2004 and 2014 has a smaller contribution in explaining the in- (when income inequality was falling), more slow- equality change over time. ly between 2014 and 2019 (when income inequal- 25. A Shapley decomposition gives the “marginal effect on [the indicator] of eliminating each of the contributory factors in sequence, and then assigns to each factor the average of its marginal contributions in all possible elimination sequences. This procedure yields an exact additive decomposition of [the indicator] into [the number of] contributions” (Shorrocks 2013, 3). 26. It is not possible to completely distinguish public transfers from private transfers in the data between 2004 and 2012. Between 2012 and 2014, public transfers reduced inequality while private transfers increased it. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 26 F I G U R E 2 .1 9 Average per capita monthly income in 2017 (RM) TREND IN MEAN 2,500 INCOMES PER C A P I TA B Y Kuala Lumpur 2,000 S TAT E , 2 0 0 4 – 2 2 Putrajaya Selangor 1,500 Sarawak Kelantan 1,000 Sabah 500 Source: World Bank staff calculations based 2004 2007 2009 2012 2014 2016 2019 2022 on DOSM HIES/BA. Mean incomes across states have not con- While gaps across states remain, there is sig- verged over time; mean incomes in the East nificant variation within them. The contribu- states and in Kelantan lag behind. There have tion of interstate inequality, which reflects mean long been concerns regarding the spatial distribu- state incomes, to overall inequality has remained tion of Malaysia’s economic success. For instance, at less than 15 percent over time (table 2.1).28 before the 1997 Asian Financial crisis, the 10-year Meanwhile, intrastate inequality is the largest average national growth rate was 9 percent, but it contributor to national inequality. Until 2016, the was only 4 percent in Sabah (World Bank 2022a). sampling for the HIES/BA did not allow produc- At the household income level and for the peri- ing estimates (for example, of poverty or average od analyzed in this report, these patterns are also incomes) disaggregated at the district level to en- evident. Incomes have grown in all states since able an understanding of intrastate inequality,29 2004, but average incomes in the richest states and no official poverty maps30 for Malaysia exist have grown faster than the average incomes in as well, but some level of understanding of spa- the poorest states (figure 2.19) and relative gaps tial inequality at a lower level than the states is have remained and even grown slightly: the mean possible using a global data set of a measure of income per capita in Kuala Lumpur, the capital a relative standard of living produced using con- and the richest territory,27 was 2.5 times that in nectivity data, satellite imagery, and other non- Sabah, the poorest state, in 2004 and 2.7 times traditional data sources (Blumenstock et al. 2015) in 2019. Average household incomes in Sabah, The measures for the “relative wealth index” at Kelantan, and Sarawak remained below those the district level are shown in figure 2.20; the in other states, whereas those in Kuala Lumpur, measures show that, while overall the states in Putrajaya, and Selangor remained on top and sig- East Malaysia (Sabah and Sarawak) are less well- nificantly above the mean incomes in the rest of off, pockets of poverty exist within richer states the country. and poorer states have pockets of prosperity. 27. Kuala Lumpur is a federal territory. 28. Using an ELMO specification of the mean log deviation (MLD) as the inequality measure. We use the MLD instead of the Gini since the Gini is not additively decomposable into intra- and inter-group inequality. 29. HIS is representative at the district level since 2016 and HES since 2019. 30. Estimates that combine census data and household survey data to produce subnationally disaggregated measures of poverty at the level of a small area. 27 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA TA B L E 2 . 1 S H A R E O F I N E Q U A L I T Y AT T R I B U TA B L E T O THE BET WEEN-GROUP COMPONENT (%)   2004 2007 2009 2012 2014 2016 2019 2022 Gini 45.1 43.6 44.4 42.4 40.1 39.5 39.8 39.0 MLD 35.2 32.8 34.0 30.8 27.1 26.2 26.7 25.6 (a) State, 16 Intergroup share, % 14.3 11.7 11.9 10.8 11.0 10.7 12.0 14.2 (b) Ethnic, 3 Intergroup share, % 14.8 13.1 8.7 10.8 10.1 11.1 12.6 12.8 (c) State-Ethnic, 48 Intergroup share, % 22.4 19.5 17.1 15.7 16.8 16.4 17.7 19.9 Note: ELMO decomposition of total inequality (mean log deviation, MLD) reported. ELMO decomposition accounts for differences in the number of groups. In 2007, 2009, and 2014, smaller groups were removed for the State-Ethnic decomposition to achieve conver- gence in the code. FIGURE 2.20 R E L AT I V E W E A LT H I N D E X ( P R O J E C T I O N T O 1 0 0 M P O P U L AT E D A R E A S ) Source: World Bank staff calculations based on http://www.povertymaps.net/. Notes: Relative wealth indexes, produced using deidentified connectivity data, satellite imagery, and other nontraditional data sources, are microestimates of the relative standard of living within countries (see Blumenstock et al. 2015; Chi et al. 2022.) Ethnicity has traditionally been considered the poorest Bumiputera have grown faster than another salient domain of inequality in Ma- the incomes of the richest Bumiputera) and has laysia. The Bumiputera, who are the majori- also been more pro-poor than income growth for ty of the population but also historically the other ethnicities (steeper GICs for the Bumiput- poorest group, have experienced more rapid era than for the Indian and Chinese) (figure 2.22). pro-poor growth. The Bumiputera constitute Further, for the most part, income growth for the close to 70 percent of the population and hold Bumiputera has been higher than income growth 62 percent of the income (figure 2.21), which is for other ethnic groups in the same decile of the a small improvement since 2004, when they ac- national income distribution. Only the Bumiput- counted for 67 percent of the population and 55 era in the richest 20 percent of the distribution percent of the income. Over 2004–22, income experienced less income growth than other eth- growth among the Bumiputera has been pro- nic groups. poor within the ethnic group (the incomes of A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 28 FIGURE 2.21 100 INCOME SHARES 90 Income shares by ethnic groups (percent) BY ETHNIC 80 G R O U P, 2 0 0 4 – 2 2 70 60 50 40 30 20 10 0 Source: World Bank 2004 2007 2009 2012 2014 2016 2019 2022 staff calculations based on data from DOSM Bumiputera Chinese Indian HIES/BA. FIGURE 2.22 8 GROW TH Annual growth of ethnic deciles, % INCIDENCE 6 CURVES BY E T H N I C I T Y, 2004–22 4 2 0 Source: World Bank 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 staff calculations National income percentile based on data from DOSM HIES/BA. Bumiputera Chinese Indian Despite much progress, gaps between the vantaged group. The contribution of the inter- Bumiputera and other ethnic groups persist. ethnic inequality to overall inequality is relatively Pro-poor growth helped narrow the income gaps small; differences in average incomes across eth- across ethnic groups, but only by small amounts, nic groups have fallen slightly over time, from 15 partly due to a much lower starting point for the percent to 13 percent of the total inequality (table Bumiputera. In 2004, the average income of the 2.1). On the other hand, when making groups that Chinese was 1.9 times the income of the average combine ethnicity and region, the contribution Bumiputera; in 2022, this ratio had declined to 1.8 of intergroup inequality is 20 percent,31 which times (figure 2.23). A similar pattern is observed means this is the most important source of av- when looking at the PG by ethnic grouping. In erage differences across groups. The Bumiputera 2004, the income of the Bumiputera had to grow in Sabah and Sarawak have the lowest mean in- on average by over five times to reach the HIC come of all subgroups of ethnicities and regions equivalent, compared with four times for the (East Malaysia and Peninsula) (figure 2.25).32 In Indian and less than three times for the Chinese contrast, Chinese and especially Indians in East (figure 2.24). By 2022, the factor had declined to Malaysia fare much better. Further, while Bumi- two times for the Bumiputera, but it also declined putera in the Peninsula are the second-poorest for the other ethnic groupings. group on average, their incomes are similar to those of the Indians in the Peninsula or the Chi- Interethnic inequality is still important. Bu- nese in East Malaysia. miputera in East Malaysia are the most disad- 31. This decomposition takes into account that the number of groups is larger when using region and ethnicity than when using either of those on their own. 32. While the HIES/BA is not designed to provide representative estimates at the ethnic X region level, the sample sizes for these subgroups are large enough to produce reasonable estimates. 29 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 2.23 2500 MEAN INCOME BY ETHNIC G R O U P, 2 0 0 4 – 2 2 2000 Mean per capita income 1500 1000 500 Source: World Bank 2004 2007 2009 2012 2014 2016 2019 2022 staff calculations based on data from DOSM Chinese Indian Bumiputera HIES/BA. FIGURE 2.24 6.0 PROSPERIT Y GAP TRENDS BY 5.0 E T H N I C G R O U P, 4.0 2004–22 Prosperity gap 3.0 2.0 1.0 Source: World Bank 0.0 staff calculations 2004 2007 2009 2012 2014 2016 2019 2022 based on DOSM HIES/BA. Bumiputera National Indian Chinese FIGURE 2.25 2,000 MEAN INCOME Mean per capita monthly income BY ETHNIC GROUP AND 1,500 REGION, 2004–22 1,000 500 2004 2007 2009 2012 2014 2016 2019 2022 Source: World Bank staff calculations Chinese, East Bumiputera, Peninsular based on DOSM HIES/BA. Indian, Peninsular Bumiputera, East A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 30 B O X 2 .1 H O U S E H O L D D E M O G R A P H I C S A N D I N E Q U A L I T Y Household demographics can affect inequality today and in the future. Households spend their income on either current consumption or saving for future investments. But if richer households are smaller and have fewer dependents, as is often the case in developing countries, the total income of the household needs to finance the consumption of fewer members; this makes the per person current income larger and also leaves a larger share to save or invest. Malaysian households have been shrinking fast, but household sizes have not fallen at the same pace across the income distribution. Fertility rates have fallen fast in Malaysia; from over six births per woman in the early 1960s to three in the late 1990s and early 2000s, to 1.8 today. Fertility declined especially rapidly in the 1970s but also in the late 1990s to early 2000s. This decline in fertility is reflected in falling household sizes. In the 2000 Population Census, the average household size was 4.6 members; this fell to 4.2 by 2010 and further to 3.8 in 2020. Over the period of analysis in this report (2004–22), household size among poorer households fell by 10 percent in the first decade (2004–14) and by 7 percent over the second time period (2014–22). In contrast, while richer households were also getting smaller, they were shrinking slower than poorer households in the first decade but shrunk faster in the second decade (7 and 11 percent, respectively). The rapid decline in household size, especially among the poor, contributed to the decline in inequality in the first decade. Between 2004 and 2014, the average household size fell by 5.3 percent. The decline was much faster for households in the poorest 40 percent of the distribution and even more so for the poorest 10 percent, for which the average household size fell by almost 10 percent. Because poorer households had fewer and fewer people to spread their incomes, the per person income was rising; this contributed to the decline in in- equality observed during this period. The Gini would have been half a point higher (40.5) in 2014 had household sizes stayed the same as they were in 2004, instead of declining significantly among poorer households. F I G U R E B 2 .1 .1 F I G U R E B 2 .1. 2 CHANGE IN HOUSEHOLD SIZE ACROSS THE INEQUALIT Y FOR DIFFERENT FERTILIT Y INCOME DISTRIBUTION, 2004–22 SCENARIOS, 2004, 2014, 2022 0% 46 -5% Change in household size 44 -10% Gini 42 -15% 40 -20% 1 2 3 4 5 6 7 8 9 10 Decile 38 2004 2014 2022 2004-14 2014-22 Actual Gini Counterfactual (faster decline in household size) Counterfactual Counterfactual (unchanged (decline in household size) household size as in previous decade) Source: World Bank staff calculations based on DOSM HIES/BA. Note: The average household size was estimated for every per capita income percentile in 2004, 2014, and 2022. A baseline Gini was estimated for 2004 and 2014 based on the total household income divided by the average size for the relevant percentile (rather than the actual household size). A first counterfactual Gini for 2014 (or 2022) was estimated based on the total household income divided by the average size for the relevant percentile in 2004 (or 2014) (i.e., what would have happened had demograph- ics remained constant). A second counterfactual Gini for 2022 was estimated based on the total household income divided by the percentile average size in 2004 multiplied by the change in the percentile average household size between 2004 and 2014. A third counterfactual Gini for 2022 was estimated based on the total household income divided by the percentile average size in 2004 multiplied by the double of the change in the percentile average household size between 2004 and 2014. The difference between the baseline and counterfactual scenarios was then applied to the actual 2014 (or 2022) Gini. 31 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA After 2014, the decline in household size was more even across the distribution; only the households in the poorest 20 percent shrank at a slower pace. Over 2014–22, household size fell by about 10–11 percent for most households across the income distribution. Only households in the bottom two deciles had lower declines in size (7 and 8 percent over this period), while those in deciles 5 and 6 had the fastest declines (12 and 13 percent). Had household sizes continued to decline at the same rate as in the previous period—at lower rates overall but declining faster among poorer households—the Gini would have fallen to 38.5 instead of 38.9. Had household siz- es instead declined by roughly the same average rate as they did between 2014 and 2022 but faster among poorer than richer households (as in 2004–14), then inequality would have been even lower, reaching 38.2. Malaysia has been aging rapidly, and the trend is only expected to increase. The proportion of the popula- tion aged 60 years and older is projected to increase to 20 percent by 2040, from 10 percent in 2020 (Department of Statistics Malaysia projections based on census data). People living longer can potentially increase household sizes (although smaller nuclear family arrangements have become more popular over time in Malaysia), but, un- like children, older individuals can also contribute to household income as well as to care and domestic work responsibilities. Nonetheless, in Malaysia, work declines rapidly with age. According to the Asian Development Bank aging survey (ADB 2023), the proportion of 60- to 69-year-olds still working is 28 percent, compared with 69 percent among 40- to 49-year-olds. Retirement, but also ill health or disability, and, to a lesser extent, house- hold care commitments, are the main reasons for those older than 60 not working. Aging can increase inequality due to the cumulative effects of random shocks to income and wealth. A lifecycle model of savings predicts that as a cohort ages, random shocks to income and wealth accumulate and increase inequality (Deaton and Paxon 1995). This has been the case in Taiwan and mainland China, where the levels of inequality are indeed higher for older cohorts (Deaton and Paxson 1995; Hanewald, Jia, and Liu 2021), although the same has not been the observation consistently in other countries, for example, the United States (Hungerford 2020). We examine whether this is the case in Malaysia, using the synthetic panel data from Rongen and Lanjouw (2024) for income recipients in urban areas of the country and defining five-year birth cohorts. Inequality in incomes (note this is gross income) for this exercise is measured by the mean log deviation, which is decomposed into intercohort and intracohort components. Inequality is higher in older cohorts, and there is also some evidence of increasing intracohort inequal- ity in Malaysia. Inequality measures are higher for older cohorts. Intracohort inequality is higher for younger co- horts (born in 1970 and after), and in more recent years (2016 and after). Recent increases in intracohort inequality appear to be driven by incomes at the lower parts of the distribution falling farther away from the mean income. However, these patterns diminish once household composition is accounted for, because we observe different intracohort inequality patterns for per capita incomes compared with those described for income earners. F I G U R E B 2 .1 . 3 I N E Q U A L I T Y ( M E A N L O G D E V I AT I O N ) W I T H I N B I R T H C O H O R T S , 2 0 0 5 – 2 0 .6 .6 .6 .5 .5 .5 .4 .4 .4 .3 .3 .3 .2 .2 .2 .1 .1 .1 2005 2010 2015 2020 2005 2010 2015 2020 2005 2010 2015 2020 Survey year Survey year Survey year 1945 1950 1955 1960 1965 1970 1975 1980 1985 Source: Rongen and Lanjouw (2024) based on DOSM HIES/BA. Note: Each line represents the development of inequality within a five-year birth cohort. The estimates are based on income recipients in urban areas. The measure of inequality is mean log deviation. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 32 What would inequality have been in 2022 if Malaysia were already an aged or super-aged society? A counterfactual analysis shows what would happen if the average incomes for different age groups are kept as they are in 2022 but their population shares were changed to reflect the projected age composition in 2043 and 2056.33 Under these scenarios, inequality among income recipients would be significantly higher: an increase of 5.4 percent based on the population shares in 2043, and up 8.5 percent using the shares in 2056. The demographic transition in Malaysia—falling fertility rates and increasing life expectancy—has had a small but significant impact on changes in inequality over time. On the other hand, population aging is likely to have an increased impact on household demographics and inequality in the nearby future. Giv- en that Malaysia has only recently become an aging society, implications for inequality may not yet be discernible but are likely to increase in the future. 33. UN-projected population shares show Malaysia at the threshold of becoming an aged (14 percent of the population 65 or above, in 2043) and super-aged (20 percent in that group, in 2056) society. The UN-projected shares are as a percentage of the full population. These projections are adjusted to reflect the share of the income-receiving population. The analysis is based on the population 15 years or older, divided into the age groups 15–24, 25–49, 50–64, and 65+. 33 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 03 ECONOMIC MOBILITY TRENDS A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 34 3.1 ECONOMIC MOBILIT Y TRENDS T his section focuses on mo- analysis is based on a predicted income,34 which bility in and out of pover- is at the core of the synthetic panel, it results in ty and economic security upper- and lower-bound estimates for absolute over time. In the absence mobility derived from two extreme cases of the of real panel data for Ma- assumed relationship between the error terms laysia, the study of mobil- of the income regressions in Rounds 1 and 2 of ity relies on synthetic pan- the surveys. For the analysis of absolute mobility, els. The construction of synthetic panels exploits the population is split into three classes (poor35/ multiple rounds of comparable cross-section vulnerable/economically secure class). In this re- data and applies the methodology of Dang et al. gard, the new analysis departs from the Malaysian (2014). Synthetic panels for Malaysia were first categories traditionally used to split the popula- built by Rongen et al. (2024), and the work of Ron- tion (B40/M40/T20) by instead conceptualizing gen and Lanjouw (2024) for this report extends groups based on the notions of vulnerability to the period of analysis to include the most recent poverty and economic security (see box 3.1). survey data from 2019 and 2022. Because this B O X 3 . 1 I N C O M E G R O U P S A N D E C O N O M I C S E C U R I T Y C L A S S E S I N M A L AY S I A Traditionally, Malaysia has used three categories to split the population according to income level: the top 20 percent, middle 40 percent, and bottom 40 percent (T20/M40/B40). B40 is the group targeted by social assis- tance policies. The thresholds to define these groups are set for every survey year relative to the distribution of total household income. For instance, in 2019, these were below RM 4,850 for B40; RM 4,850 to RM 10,959 for M40; and above RM 10,959 for T20.36 Instead, this report constructs household categories that are grounded in the concepts of vulnerability to poverty and economic security and developed for the Malaysian context. The estimations use data from the 2019 House- hold Income, Expenditure and Basic Amenities Survey (baseline pre-COVID data), and the monetary value of the thresholds derived for this year are held constant over time in real terms for other years. The core of the meth- odology follows the approach of Chaudhuri (2003) to estimate the likelihood of falling in poverty when panel data relying on the cross-sectional variability in income are absent. Then, following the work of López-Calva and Ortiz-Juárez (2014) in Latin America and the Caribbean and World Bank (2016a) in Indonesia, this approach is extended to derive economically secure class lines. Current results use the Malaysia household-specific poverty lines and per capita net household income (see the appendix for the differences of this approach with respect to the official poverty measurement). The threshold to define the vulnerable is set at an income level equivalent to a probability of 20 percent or more of falling into poverty. Analogously, the threshold to define the economically secure is set at an income equivalent to a probability of 20 percent or more of being vulnerable. Given Malaysia’s high-income country (HIC) aspirations, the upper-class line is set at the household equivalent to the HIC thresh- old as used in the prosperity gap (PG) calculations. The results reveal that B40 and M40 are not homogenous groups (figure B3.1.1). For instance, B40 includes the poor and the vulnerable, but also 20 percent of the Malaysian population who, while not yet economically secure, are neither poor nor vulnerable (they are called the “aspiring” class). Moreover, while the poor and vulnerable would benefit from social assistance policies to improve their living conditions and raise their incomes, the as- piring class may benefit more from policies that may allow them to save and invest in riskier but also potentially more rewarding economic activities. Similarly, M40 is heterogeneous, containing most of the economically se- cure, but also some of the upper class. 34. The idea is to predict the Round 1 income for the households in the Round 2 survey by applying the coefficient vector resulting from the Round 1 income regression ( β_1 ) to the characteristics of the households in Round 2 ( x_i2). 35. The poverty threshold can be set in a number of ways; however, to simplify the presentation, this report focuses on the results using a poverty line set at the 40th percentile of the income distribution at the beginning of the period (2004), which is equivalent to RM 527 (in 2016 prices) per capita per month. Another option, which results in a very similar line, is to use the per capita average of the household-specific poverty lines calculated by the Department of Statistics Malaysia (DOSM) for the households in the 2019 sample, at RM 559 per capita per month. A third option is to follow the same approach but using the 2022 household-specific poverty lines, resulting in a threshold of RM 618 per capita per month. The results are qualitatively the same, irrespective of which of the three poverty lines is used (see Rongen and Lanjouw 2024). 36. For 2022, the threshold for B40 was RM 5,249; for M40 was between RM 5,250 and RM 11,819; and for T20 was above RM 11,819. 35 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE B3.1.1 B40/M40/T20 AND ECONOMIC SECURIT Y CL A SSES 100 90 Upper class 80 70 60 Economically secure class Percent 50 40 Aspiring 30 20 Vulnerable 10 Poor 0 MYS groups Economic (household income) security approach T20 M40 B40 Source: World Bank staff calculations based on DOSM HIES/BA. Note: B40 = bottom 40 percent; M40 = middle 40 percent; T20 = top 20 percent. Significant upward mobility out of poverty in chronic poverty nationally and for all groups, as means that large shares of the population will be shown next, has coincided with the slow- exited chronic poverty over two decades and down in national inequality reduction after 2014. many were able to live out of poverty year to year. Absolute mobility in income is captured Chronic poverty has fallen markedly for the by the proportion of people who have moved Bumiputera, especially among those in the to/from an income class that (1) stayed poor, Peninsula. By ethnic group, the pattern is simi- (2) moved up from or down into poverty, or (3) lar to national trends, with chronic poverty fall- stayed above poverty.37 Nationally, there has ing especially markedly for the Bumiputera. In been a large increase in the share of the popula- 2004–07, chronic poverty rates were high (30–35 tion that is persistently out of poverty, reaching percent), somewhat lower for the Indian Malay- close to 90 percent in the 2019–22 interval. At sian and 6–10 percent for the Chinese Malaysian. the same time, the share of the population that By 2019–22, chronic poverty was below 5 percent experienced downward mobility into poverty and approaching the rates for other ethnicities decreased through the period and chronic pov- (figure 3.1). Among all ethnic groups, the Penin- erty fell dramatically from 22–28 percent of peo- sular Bumiputera saw the largest increase in stay- ple in 2004–07 to under 2–3 percent in 2019–22. ing out of poverty, while the Bumiputera in East Upward mobility out of poverty was the highest Malaysia retained the highest rates of chronic between the 2009–12 and 2012–14 intervals but poverty, despite improvements over time (figure slowed down since then. These patterns are par- 3.2). There was also convergence across geog- tially related to ever smaller chances of poverty raphies; chronic poverty reduced especially for as the number of chronic poor has declined sig- rural populations, but pockets of chronic poverty nificantly. Further, the slowdown in the declines remained in rural areas in East Malaysia. 37. The synthetic panels allow assessing two main outcomes: (1) joint probabilities, or the probabilities of belonging to the same income class (e.g., poor) in two time periods, and (2) conditional probabilities, or the likelihood of moving out of a group, given being in the group in the first time period. Here we focus mostly on the description of the first set of results; the remaining analy- sis can be found in the accompanying background paper (Rongen and Lanjouw 2024). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 36 F I G U R E 3 .1 40 40 CHRONIC POVERT Y BY Share within the subgroups Share within the subgroups ETHNIC GROUP 30 30 20 20 10 10 Source: Rongen and Lanjouw 2024. Note: Upper and 0 0 lower bounds, point 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 estimates, and two standard deviation ranges for poverty (poverty line at RM 527). Households below Range Bumiputera Range Bumiputera the poverty line in two Range Chinese Malaysian Range Indian Malaysian consecutive survey waves. Range indicates the two standard deviation band around the point estimate FIGURE 3.2 Chronic poverty CHRONIC Peninsular Bumiputera East Malaysia Bumiputera POVERT Y 60 WITHIN THE BUMIPUTERA Share within the subgroups 40 20 Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds, point 0 estimates, and two standard deviation 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 ranges for poverty (poverty line at RM 527). Households below the poverty line in two consecutive survey Bounds Point estimate +/- two standard deviations waves. There has been an increase in national eco- period, but so was downward mobility, which did nomic security, which at least doubled in not decrease over time. two decades. Mobility analysis can also be made focusing on movements in the middle of the The Bumiputera and people in East and rural income distribution—using higher thresholds Malaysia still lag in persistently reaching the instead of a poverty line to define the income economically secure class. The different ethnic classes.38 As with poverty, there was also an in- groups have not reached persistent economic se- crease in the share of the population persistently curity at the same rate. For example, in 2019–22, belonging in the higher classes (economic secure 56–60 percent of the Chinese were persistently or upper classes); the share reached 31–36 per- above the economically secure class line, where- cent in the 2019–22 interval from 8–11 percent in as this held for approximately 33–36 percent of 2004–07. Upward mobility into the economical- the Indian and only 24–28 percent of the Bumi- ly secure class was stable across the entire study putera (figure 3.3a). Subnationally, urban and rural 38. Again, there are two scenarios: (1) using a poverty line of RM 527 and a vulnerability line of RM 1,120, which is estimated by applying the method proposed by Dang and Lanjouw (2017) to the 2004–07 transition interval and selecting a vulnerability index of 0.33; and (2) using the methods of Chaudhuri (2003) and López-Calva and Ortiz-Juárez (2014) to define vulnerability and economically secure classes (box 3.1) and using a 10 percent probability of future poverty; this yields a vulnerability line of RM 920 (in 2016 prices) and an “economically secure class line” of RM 1,590. The discussion here focuses on the second scenario, and the results for the first scenario are discussed in detail in the accompanying background paper. 37 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA areas, as well as East Malaysia and the Peninsular Peninsular Malaysia and the absolute gap with regions, have all seen increased economic securi- rural East Malaysia has grown over time (figure ty, but there are gaps. Persistent economically se- 3.3b). The Bumiputera in East Malaysia have seen cure class status is much more common in urban less upward mobility than other groups. FIGURE 3.3 ABSOLUTE INCOME MOBILIT Y BY ETHNICIT Y ( E C O N O M I C A L LY S E C U R E C L A S S ) Upward mobile Bumiputera Chinese Malaysian Indian Malaysian 50 Share within the subgroups 40 30 20 10 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Downward mobile Bumiputera Chinese Malaysian Indian Malaysian 50 Share within the subgroups 40 30 20 10 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Secure in both periods Bumiputera Chinese Malaysian Indian Malaysian 80 Share within the subgroups 60 40 20 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds of shares per transition category. Households that are economically secure over two consecutive survey waves. Upper and lower bounds of shares per transition category. Economically secure class line of RM 1,590. This group excludes the poor and vulnerable groups, based on applying the method of Chaudhuri (2003) and López-Calva and Ortiz-Juárez (2014) and a 10 percent probability of falling into poverty to define the vulnerable. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 38 There has also been increasing downward This seems to be the case after 2014 for rural ar- mobility from economic security for some eas of East Malaysia and for the Bumiputera in disadvantaged subgroups. There are some peo- the East (figures 3.4 and 3.5). ple for whom downward mobility has increased. F I G U R E 3 .4 ABSOLUTE INCOME MOBILIT Y BY RURAL/URBAN AND PENINSUL AR M A L AY S I A / E A S T M A L AY S I A ( E C O N O M I C A L LY S E C U R E C L A S S ) Upward mobile Downward mobile Secure in both periods 50 Urban PM Rural EM 50 Urban PM Rural EM Urban PM Rural EM 50 Share within the subgroups Share within the subgroups Share within the subgroups 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Source: Rongen and Lanjouw (2024). Note: Upper and lower bounds of shares per transition category. Households that are economically secure over two consecutive survey waves. Upper and lower bounds of shares per transition category. Economically secure class line of RM 1,590. This group excludes the poor and vulnerable groups, based on applying the methods of Chaudhuri (2003) and López-Calva and Ortiz-Juárez (2014) and a 10 per- cent probability of falling into poverty to define the vulnerable. EM = East Malaysia; PM = Peninsular Malaysia. F I G U R E 3 .5 ABSOLUTE INCOME MOBILIT Y BY ETHNICIT Y AND PENINSUL AR M A L AY S I A / E A S T M A L AY S I A ( E C O N O M I C A L LY S E C U R E C L A S S ) Upward mobile Downward mobile Secure in both periods PM Bumiputera EM Bumiputera PM Bumiputera EM Bumiputera PM Bumiputera EM Bumiputera 50 50 50 Share within the subgroups Share within the subgroups Share within the subgroups 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Source: Rongen and Lanjouw (2024). Note: Upper and lower bounds of shares per transition category. Households that are economically secure over two consecutive survey waves. Upper and lower bounds of shares per transition category. Economically secure class line of RM 1,590. This group excludes the poor and vulnerable groups, based on applying the methods of Chaudhuri (2003) and López-Calva and Ortiz-Juárez (2014) and a 10 per- cent probability of falling into poverty to define the vulnerable. EM = East Malaysia; PM = Peninsular Malaysia. 39 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA The national pattern of upward absolute expressed that they were better off than their mobility is consistent with people’s percep- parents. However, the feeling of upward mobili- tions of an increased income status relative ty becomes much lower among those who think to their parents. On average, about 60 percent of themselves as belonging to the lower quintile, of respondents to the Malaysian perception sur- but increases with better perceived welfare status vey, nationally and across different ethnic groups, (figure 3.6). FIGURE 3.6 PERCEIVED INCOME MOBILIT Y (HE AD OF HOUSEHOLD V I S -À- V I S PA R E N T S ) B Y P E R C E I V E D I N C O M E C L A S S 100 90 Perceived income mobility (percent) 80 70 60 50 40 30 20 10 0 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Higher Same Lower Source: DOSM–World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: If we divide all households in Malaysia into 10 equal groups, with 1 being the poorest group and 10 being the richest group, which income class would your household belong to? If we divide all households in Malaysia into 10 equal groups, with 1 being the poorest group and 10 being the richest group, which income class would your parents’ household belong to when they were at your age? (1 is the lowest or poorest income class, while 10 is the highest or richest income class). Quintiles are perceived quintiles based on the income class that a respondent felt their household belongs to. “Lower/Same/Higher” if a respondent felt their household income was “Lower/Same/Higher” than their parents’ income. FIGURE 3. 7 DIFFERENCE BET WEEN ACTUAL AND PERCEIVED INCOME MOBILIT Y FOR CHILDREN 35 30 Percent of populations 25 20 15 10 5 0 1 2 3 4 5 Expected children quintile Quintile of parent’s income Q1 Q2 Q3 Q4 Q5 Difference Source: DOSM–World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: If we divide all households in Malaysia into 10 equal groups, with 1 being the poorest group and 10 being the richest group, which income class would your household belong to? If the current situation continues, which income group do you think your children or younger members of your family will belong to when they reach your age? (1 is the lowest or poorest income class, while 10 is the highest or wealthiest income class).Quintiles are perceived quintiles based on the income class that a respondent felt their household belongs to. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 40 People’s expectations of economic mobil- reflecting a positive difference between the actu- ity for future generations are optimistic, al and perceived mobility, while very few expect in some cases overly so. If respondents were that their children stay at the bottom. This may ranked according to their actual income, nearly seem contradictory to the sense of a widening 70 percent of them thought that their children income gap in the country, but in fact, it is not would be somewhat better off than their current uncommon to be more optimistic about one’s welfare. However, people’s expectations on their circumstances (and one’s children by extension) children’s future welfare do not appear correlated than about the state of the country as a whole.39 with their current welfare; most people expect their children to be in the middle of the actual But poorer people have limited expectations income distribution (Quintile 3 or Quintile 4). about how much improvement their children Interestingly, Chinese respondents tend to be can achieve. People who think they are in the less optimistic than the Bumiputra or the Indian poorest deciles of the income distribution have Malaysians. Meanwhile, people’s expectations limited expectations about how much improve- about how much their children’s future welfare ment there can be with respect to their children’s can improve depends on their perceived place mobility. At most, they expect their children to in the income distribution. Figure 3.7 shows the reach the middle of the distribution but not the difference between the actual and perceived in- top two quintiles. For instance, most people who come mobility for children. Each quintile should perceive they are in the bottom two quintiles ex- have 20 percent of the population by definition, pect their children to be just one quintile above but most people expect their children to be in the in the future, and only very few expect their chil- middle and upper-middle quintiles (Q3 and Q4), dren to “jump” to the highest quintile. People’s expectations about how much their children’s future welfare can improve depends on their perceived place in the income distribution. 39. The following are among the potential reasons (Roser and Ritchie 2018): (1) since people rarely think about grand issues such as the state of the nation or world, their response in a survey may not be well considered or even a true reflection of their beliefs; (2) differences in the framing of questions that influence an individual’s response; (3) responses are heavily influenced by recent events, but positive and slow changes do not get much attention, and, thus, people are not well informed about country progress; and (4) the fact that we feel more in direct control of our own lives but not the country’s prospects. 41 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 3 . 2 R E L AT I V E IN C OME M O BIL I T Y T his section focuses on esti- The poor are more likely to be stuck at the mates of relative mobility bottom of the distribution, whereas the rich across five quintiles in the are more likely to be secure at the top. There income distribution. The is more short-term mobility in the middle of synthetic panel is used to es- the distribution. Relative mobility patterns do timate the probability that a not change dramatically over time; approximate- household has made a partic- ly 24–32 percent of the population is upward ular transition from one quintile to another over mobile in any given interval, while the share of adjacent survey rounds. This yields a 5 x 5 tran- those moving down is generally between 27 and sition matrix across the quintiles in the income 34 percent. To enable viewing mobility patterns distribution. Deriving these estimates requires an across quintiles better, figure 3.8 shows the av- additional assumption about the correlation of erage mobility for all seven wave pairs between errors in the income model to narrow the bounds 2004–07 and 2019–22; the pattern is similar for of the synthetic panels and obtain point estimates each wave pair, however. Over half of the people of income in each survey round. In the absence of in the bottom quintile stay there in the next wave real panel data for Malaysia, income correlations (approximately over a three-year period); that are obtained from actual panel data from Colgan is, they are trapped in the bottom of the distri- (2023) and Dang and Lanjouw (2023).40 The anal- bution, improving at a slower pace than the rest. ysis here uses the smallest and largest correlation Similarly, on the other end of the distribution, half values in this set of countries and applies these to of the people in the top quintile stay at the top. In the Malaysian synthetic panel bounds to estimate, Quintiles 2, 3, and 4, there is more relative mobili- respectively, the high and low mobility bounds of ty, with movements both up and down. relative mobility; the closer the coefficient is to 1, the stickier the incomes are. FIGURE 3.8 S H O R T-T E R M R E L AT I V E M O B I L I T Y : N U M B E R O F Q U I N T I L E S MOVED UP OR DOWN 20 Percent of population changing income quintile 15 10 5 0 -5 -10 -15 Q1 Q2 Q3 Q4 Q5 -20 Quintile -4 -3 -2 -1 0 1 2 3 4 Source: World Bank staff calculations based on Rongen and Lanjouw (2024). Note: The figure shows the average mobility (number of quintiles moved up or down) for all seven wave pairs between 2004–07 and 2019–22; the pattern is similar across wave pairs, however. 40. Eleven countries in the European Union Statistics on Income and Living Conditions data set and five other developing countries. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 42 F I G U R E 3 .9 P E R S I S T E N C E AT T H E B O T T O M A N D T O P B Y R E G I O N A N D THE BUMIPUTERA ETHNICIT Y Quintile 1 in both rounds Urban Peninsular Rural East Malaysia Peninsular Bumiputera Rural East Bumiputera Share within the subgroups .4 .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Quintile 5 in both rounds Urban Peninsular Rural East Malaysia Peninsular Bumiputera Rural East Bumiputera Share within the subgroups .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Source: Rongen and Lanjouw (2024). Note: Quintiles are determined at the population level. Downward mobility among the Bumiputera Rural-East Malaysians in Sabah and Sarawak barely changes. Mobility patterns differ by eth- are more likely to be stuck at the bottom. nic group and location. The Chinese and, to a Overall mobility in East Malaysia is lower than lesser extent, the Bumiputera are more likely to in the Peninsula, especially for rural areas. Not be immobile, whereas the Indian Malaysians are only are those in rural-East Malaysia more likely increasingly more likely to be upward mobile. to be in the bottom quintile to begin with (32–38 For the Indian Malaysians, downward mobility percent), but the share of immobility in Quintile 1 across quintiles was 27–33 percent in 2004–07 increases over time for this group (to 36–42 per- and 29–34 percent in 2019–22, while immobili- cent). At the other end of the distribution, those in ty at the bottom remains constant over time, at the urban Peninsula are more likely to be at the top, 10–18 percent. but their downward mobility increases over time. 43 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 04 UNDERSTANDING THE DRIVERS OF INEQUALITY IN MALAYSIA A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 44 4.1 INEQUALIT Y FRAMEWORK: ASSETS I nequality of opportunity development has been a major factor for many represents a large share of countries that have achieved sustained economic the income inequality in growth and poverty reduction, and human cap- Malaysia. Inequality of op- ital is a key asset in the income-generating pro- portunity, which arises from cess. Malaysia’s progress on this front can be seen predetermined circumstanc- through the Human Development Index (HDI), es in life rather than from a summary measure of the average achievement individual effort or choices, represents over 60 in key dimensions of human development: a long percent of the total market income inequality and healthy life, being knowledgeable, and having in Malaysia (figure 4.1). This share has increased a decent standard of living.41 Malaysia’s score in slightly from 61 percent in 2004 to 65 percent in the HDI rose from 0.64 in 1990 to above 0.8 in 2022. Excluding education, inequality of oppor- 2016, placing the country in the category of “very tunity is lower but still accounts for approximate- high human development” in this year; but Ma- ly 40 percent of the total market income inequal- laysia continues to lag behind all countries that ity. Opportunities start to shape early in life with recently made the transition to high income (fig- the accumulation of human capital. ure 4.2). Further, improvements in human capital have been slowing and some even reversed with Malaysia has advanced rapidly in human the COVID-19 pandemic. Malaysia’s score in the development and human capital outcomes 2021 HDI fell to a level equivalent to reversing the but still lags other countries. Human capital country’s gains by five years. F I G U R E 4 .1 0.6 INEQUALIT Y OF 0.5 OPPORTUNIT Y ( AT M A R K E T 0.4 Gini coefficient INCOME) 0.3 0.2 0.1 Source: World Bank calculations 0 based on DOSM HIES/BA. 2004 2007 2009 2012 2014 2016 2019 2022 Note: Estimations based on market income Gini. The circumstances included are area of residence, state, ethnicity, edu- IOP (with education) cation of household head, and IOP (wo education) gender of household head. IOP = inequality of opportunity. Gini overall (market income) FIGURE 4.2 .9 HUMAN DE VELOPMENT INDE X AMONG HDI Score (0-1) A S P I R AT I O N A L .8 COUNTRIES, 1990–2021 .7 .6 1990 2000 2010 2020 Source: UNDP, Human Development Index (2022). Recently transitioned HICs Malaysia 41. The HDI is the geometric mean of the normalized indices for each of the three dimensions. The health dimension is assessed by life expectancy at birth. The education dimension is measured by mean of years of schooling for adults aged 25 years and more and the expected years of schooling for children of school-entering age. The standard of living dimension is measured by gross national income per capita (UNDP 2022). 45 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.3 R AT E S O F S T U N T I N G A N D L O W B I R T H W E I G H T B Y S TAT E AND ETHNICIT Y 30 Pahang Sabah & WB Other 25 Labuan Bumiputera Kedah Selangor WP KL & Putrajaya Kelantan Terengganu Melaka Stunting Malay 20 Sarawak Johor Perlis Chinese Negeri Sembilan Perak 15 Pulau Pinang Indian 10 5 8 11 14 Low birth weight State Ethnicity Source: World Bank calculations based on NHMS 2016 and 2022. Human capital accumulation starts in utero, poor cognition, and low school performance and and regional and ethnic gaps in the rates of a poor earning potential later in life. low birthweight further reflect in wider gaps in child nutrition (figure 4.3). Although ante- Access to sanitation services is an area for natal care attendance and skilled birth delivery improvement. Access to basic infrastructure, are close to universal in Malaysia,42 1 in 10 chil- which is important for healthy living conditions dren are born with low birth weight (under 2.5 ki- especially during early childhood, has grown in lograms). Slightly higher rates of low birthweight Malaysia; by 2020, virtually all of the population are found in Negeri Sembilan (11.4), Kelantan had access to electricity and less than 6 percent (14.3), and Sarawak (16.0), as well as for the Indian lacked access to safely managed drinking water (12.2) and other Bumiputera (14.5) (IPH 2016). By services. Access to sanitation services, albeit high, the time children reached five years of age, one in lags slightly; 77 percent of people used safely five children under the age of five were stunted managed sanitation services in 2020. (height-for-age Z-score [HAZ] below 2 standard deviations) (IPH 2022). This rate, which is already By the time children are 10 years old, 43 per- significantly higher than the average of 8 percent cent of them are not able to read nor under- for upper-middle-income countries (UMICs) and stand a short text. Malaysia’s enrollment rate in the East Asia and the Pacific regional average of primary education is close to universal (95 per- 8.7 percent,43 is even worse in certain areas and cent net enrollment rate in 202244), but by the age for certain ethnic groups within the country. The of 10, when children complete primary education, rates of stunting are the highest in Pahang (28 per- 43 percent of children do not reach the minimum cent) and Sabah (25 percent), as well as for other proficiency rate in reading (World Bank 2024a). Bumiputera (24 percent), but even in Kuala Lum- This means they are not able to read and under- pur and Putrajaya (23 percent), stunting is an issue stand a short text. Learning poverty in Malaysia is for urban poor families, for whom the affordabil- 8 percentage points higher than the average for ity of a quality diet is a concern (UNICEF 2018). the East Asia and Pacific region and 11 percent- Estimates for stunting rates by wealth quintile age points higher than the average for UMICs. show that stunting rates are 10–30 percent higher Learning poverty is substantially higher among than the national average among children from children from the poorest families (in the bottom poorer families (UNICEF, WHO, and World Bank quintile), at 67 percent (World Bank 2022d). 2023). Stunting can lead to poor child growth, 42. According to data from the 2022 National Health and Morbidity Survey, 98 percent of women attended four visits during preg- nancy and 98.4 percent of births were attended by skilled personnel. 43. https://data.unicef.org/resources/low-birthweight-prevalence-interactive-dashboard/. 44. http://data.uis.unesco.org/. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 46 Early learning gaps are reflected in poor quintile in reading and two levels below in math- learning outcomes at the secondary level, ematics and science (figure 4.4). This means that especially for children from the poorest fam- for the same years of education, poorer children ilies. Tests scores of Malaysian students are be- are learning less.45 hind the average test score for the (Organisation for Economic Co-operation and Development Learning outcomes worsened in the after- (OECD) countries, and children from the poor- math of the COVID-19 pandemic. Malaysia est families perform much worse. According to saw one of the longest periods of school closures 2018 results of the Programme for International in the region during the pandemic, and during Student Assessment (PISA), 15-year-old Malay- these closures, a third of the children from low-in- sian students scored worse than students from come households lacked access to online classes high-income countries in science, reading, and or mobile learning applications to continue their mathematics. Malaysian students performed the education (World Bank 2022c) (figure 4.5); this worst in reading, scoring an average of 415 points, was despite close to universal Internet access in compared with an average of 487 points scored in the county, as shown by available indicators.46 reading by students in OECD countries. Children COVID-19 widened the gap in PISA scores be- from poor families perform significantly worse on tween Malaysia and OECD countries, although test scores than children from richer families. On the score gap across socioeconomic groups re- the PISA tests, children from the lowest quintile mained stable (figure 4.4). score one level below children from the highest F I G U R E 4 .4 600 2018 600 2022 MEAN PERFORMANCE 500 500 IN PISA TESTS 400 400 BY QUARTER OF SOCIOECONOMIC 300 300 S TAT U S 200 200 100 100 0 0 Source: World Bank staff calculations based on PISA (2018, 2022). Reading Mathematics Science Reading Mathematics Science Note: OECD = Organisation for Economic Co- Bottom quarter 2nd quarter 3rd quarter Top quarter operation and Development. Overall OECD average F I G U R E 4 .5 P E R C E N TA G E 100 OF CHILDREN E N G A G E D I N AT- 80 HOME LEARNING I N T H E PA S T 60 3 0 D AY S ( M AY / J U N E 2 0 2 1 ), B Y 40 HOUSEHOLD 20 INCOME 0 Source: World Bank RM 2,000 RM 2,001 – RM 4,001 – More than 2022b. and below RM 4,000 RM 10,000 RM 10,000 45. But whether this is quality of education, lower endowments, or lower complementary investments is difficult to disentangle. 46. For instance, DOSM’s Report of ICT Use and Access by Individuals and Households Survey shows that 96 percent of the popu- lation has Internet access. A Malaysia report of the United Nations Children’s Fund (UNICEF and Centre for Justice and Crime Prevention 2020) cites a figure of 9 in 10 children aged 5 to 17 being Internet users. 47 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA There are also subnational gaps in learning and most of that variation stems from the educa- outcomes. The World Bank’s Human Capital tion component (figure 4.6). Many of the lagging Index (HCI) quantifies the level of human capi- states48 have a lower HCI than more prosperous tal that a child can expect to attain by the end of states, suggesting that limited economic prosper- secondary school, given the risks related to poor ity in those states is partly attributable to weak health and education at the time of the child’s human capital development (World Bank 2022a). birth. While the overall HCI for the country is In the lagging states and in Labuan, there are also high, using a bespoke version of this index con- quality gaps, which are reflected in lower perfor- structed for Malaysia,47 it is possible to see that mance in national tests (Sijil Pelajaran Malaysia, outcomes are lower for some states than others, SPM).49 FIGURE 4.6 M A L AY S I A- S P E C I F I C S U B N AT I O N A L H U M A N C A P I TA L I N D E X A N D S TAT E G D P P E R C A P I TA L GDP per capita (MYR million) 1.0 140,000 0.9 120,000 0.8 0.7 100,000 0.6 80,000 HCI 0.5 0.4 60,000 0.3 40,000 0.2 20,000 0.1 0.0 0 Sabah Sarawak W.P. Labuan Perlis Kedah Kelantan Perak Pulau Pinang Selangor Pahang Johor Negeri Sembilan Terengganu Melaka W.P. Kuala Lumpur + Putrajaya HCI 2019 GDP per capita (RM) B . AV E R A G E S P M G R A D E 9 8 7 6 Average grade 5 4 3 2 1 0 Sabah Sarawak Kedah Labuan Perlis Kelantan Malaysia Perak Pulau Pinang Selangor Johor Pahang Melaka Terengganu Negeri Sembilan Kuala Lumpur Putrajaya Source: World Bank 2022a. Note: SPM grades are converted into a grade point average. Lower scores indicate better perfor- mance. Weighted average grades for BM, BI, Mathematics and science for 2019. GDP = gross domestic product; RM = Malaysian ringgit; SPM = Sijil Pelajaran Malaysia. 47. Adaptation of a Malaysia-specific HCI based on the methodology of Kraay (2018). While the measurement of the survival and health dimensions of the Malaysia-specific HCI follows the measurements of the survival and health dimensions of the World Bank’s official HCI, the measurement of the education dimension has been adjusted due to data limitations. It uses the average grade point for the Malaysian Certificate of Examination for four subjects: the Malay language, the English language, mathemat- ics, and science (see World Bank 2022a). 48. Kedah, Perlis, Kelantan, Sabah, and Sarawak. 49. SPM grade is calculated as the average in Malay language, English language, mathematics, and science. Lower scores indicate better performance. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 48 Educational attainment has rapidly increased which people have outpaced their parents’ edu- in Malaysia. According to data from the Global cation (Van der Weide et al. 2021). In 2004, vir- Database on Intergenerational Mobility, 82 per- tually none of the Bumiputera had higher educa- cent of people born in Malaysia in the 1980s have tion, while in 2022, they accounted for 15 percent a higher level of education than their parents; this of the adult population, and, in the same period, ranks Malaysia at number 5 in the world in terms the share of those who had completed only lower of absolute educational mobility or the degree to secondary education halved (figure 4.7). FIGURE 4. 7 1.00 University E D U C AT I O N A L Vocational AT TA I N M E N T 0.80 and diploma OF THE BUMIPUTERA , Upper Secondary 2004–22 0.60 Lower secondary 0.40 or less 0.20 Source: World Bank staff calculations based 0.00 on DOSM HIES/BA. 2004 2007 2009 2012 2014 2016 2019 2022 FIGURE 4.8 100% 90% 80% 30% 50% 40% 20% 60% 70% 10% 0% E D U C AT I O N A L Bumiputera-Peninsula AT TA I N M E N T Bumiputera-East BY GROUP AND Chinese-Peninsula DECILE, 2022 Chinese-East Indian D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 Source: World Bank calculations based on Lower secondary or less Upper secondary DOSM HIES/BA. Vocational and Diploma University While overall educational attainment in skills, continuing further in education is challeng- Malaysia is high as it has been rising, some ing. In states where children achieve low scores children still do not advance as far if they are at the end of their primary education, the Ujian not academically ready. In 2022, one-fifth of Pencapaian Sekolah Rendah (UPSR) exam, more adults had, at the most, completed secondary ed- of young adults only complete lower secondary ucation. Among the people in the bottom decile education or less (figure 4.9). This shows how the and for the Bumiputera in East Malaysia, that gaps in human capital that start early in life build percentage rises to over 30 percent (figure 4.8). up through the life cycle. When students do not acquire the foundational 49 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.9 STUDENT PERFORMANCE (UPSR) AND Subgroup shares are based on synthetic panel estimates for ho E D U C AT I O N A L AT TA I N M E N T the first round of each interval. Quintile groups are determined Adults age 24–31 with only a lower secondary education 25.0% Sabah 20.0% Education, or less (2019) 15.0% Selangor 10.0% Perak Sarawak Kedah Johor Pulau Pinang Pahang 5.0% Kelantan Terengganu Negeri Sembilan Melaka 0.0% Kuala lumpur Perlis 55.0% 60.0% 65.0% 70.0% 75.0% 80.0% 85.0% Share of students receiving a grade C or below in math, UPSR, 2017 Source: World Bank 2024a. Note: UPSR = Ujian Pencapaian Sekolah Rendah. Quality of education is essential for improv- est families and in the lagging states, are learning ing the productivity of workers and acquir- at school will negatively affect their ability to earn ing the right skills. Adjusting for what children a good return on their education. This example actually learn in school, the expected years of also shows that, despite high levels of access, school for Malaysia is only 8.9 years, compared low-quality services might be leaving many peo- with 12.5 unadjusted expected years of school ple disconnected from opportunities. More sys- (World Bank 2024a). The limitations in what Ma- tematic information on service quality needs to laysian children, especially those from the poor- be collected to understand these gaps better. The limitations in what Malaysian children, especially those from the poorest families and in the lagging states, are learning at school will negatively affect their ability to earn a good return on their education. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 50 4.2 INEQUALIT Y FRAMEWORK: RETURNS TO ASSETS IN THE LABOR MARKET E mployment is the main bution (figure 4.10). Taxes and transfers represent source of income for Ma- roughly -10 percent of household income overall, laysian households; thus, but public transfers have positive contributions the dynamics of the labor to income even for the top deciles of the distri- markets and the distribu- bution. Inequality in employment income, both tion of labor incomes are salaries and income from self-employment, is essential to understand higher than the overall inequality, but because inequality. Salaries and wages alone represent salaries and income from self-employment have 50–60 percent of household income, with their a large contribution to the total income, the evo- share declining over time, and self-employment lution of the overall Gini over time follows rough- income has a share close to another 40 percent ly the evolution of the employment income Gini in household income (table 4.1). The share of em- (figure 4.11). ployment income rises along the income distri- TA B L E 4 . 1 O M E S Oshares I N C Subgroup URCE A S A on S based are SH F estimates for household heads aged between 25 and 60 in A R E Opanel synthetic T O TA L first the INC round OME , each of 2 0 0interval. 4 – 2 2 Quintile groups are determined at population level. Income source 2004 2007 2009 2012 2014 2016 2019 2019 2022 Subgroup shares are based on synthetic panel estimates for household heads aged between 25 and 60 in the first round of each interval. Quintile groups are determined at population level. Wages + Salary 60 46 59 56 55 54 52 54 110% Self-employment 39 29 37 40 39 90% 36 39 37 2019 Property 11 8 13 11 13 70% 15 15 14 110% 50% Taxes and transfers  -11 16 -9 -8 -7 -5 -6 -5 90% 30% p shares are based on synthetic panel estimates for household heads aged between 25 and 60 in Source: ound of each interval. World Quintile Bank are staff groups calculations determined atbased on DOSM population HIES/BA. 70% level. 10% 50% -10% FIGURE 4.10 2019 30% -30% 10% 1 2 3 4 5 6 7 8 9 10 INCOME 110% -10% SOURCES AS A Property Public transfers 90% -30% S H A R E O F T O TA L 70% 1 2 3 4 5 6 7 employment Self 8 9 10 Transfers INCOME BY 50% Wages Taxes INCOME DECILE, 2019 30% Property Public transfers 10% Self employment Transfers -10% Wages Taxes Source: World Bank -30% staff calculations based 1 2 3 4 5 6 7 8 9 10 on DOSM HIES/BA. Property Public transfers FIGURE 4.11 60 Self employment Transfers Wages Taxes GINI FOR 55 EMPLOYMENT 50 INCOME PER Gini index C A P I TA , 2 0 0 4 – 2 2 45 40 Source: World Bank staff 35 calculations based on DOSM HIES/BA. 30 Note: Employment income includes wages and income 2004 2007 2009 2012 2014 2016 2019 2022 from self-employment. 51 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.12 6.0 Annual income growth, % GROW TH 5.0 INCIDENCE CURVE FOR EMPLOYMENT 4.0 INCOME PER 3.0 C A P I TA , 2 0 0 4 – 2 2 2.0 Source: World Bank staff 1.0 calculations based on DOSM HIES/BA. 1 2 3 4 5 Note: Employment income Total per capita household labor income quintile includes wages and income from self-employment. 2004–2014 2014–2019 2019–2022 Poorer households tend to be engaged in clined as a share of employment, but mostly be- agriculture or providing low-value services, cause of a reduction in the upper-middle and top whereas richer households are more likely deciles. Many of the workers in the bottom and to be engaged in manufacturing or providing middle deciles shifted from agriculture and man- high-value services. Although agriculture has ufacturing to low-value-added services. In con- declined substantially over time, it still employs trast, workers at the top of the household income close to 20 percent of workers at the bottom of distribution shifted to high-value-added services the household income distribution (figure 4.13), (mostly related to professional activities), which and workers in the lagging states of the East (17 now employ over 40 percent of workers in the percent in Sabah and 14 percent in Sarawak in top 20 percent of the income distribution, com- 2022), but very few workers at the middle and the pared with about 25 percent in the bottom 40 top. Employment in manufacturing has also de- percent. F I G U R E 4 .1 3 SECTOR OF EMPLOYMENT BY SECTOR AND DECILE, 2004, 2014, AND 2022 2004 2014 2022 100% 90% 80% Percent of workers 70% 60% 50% 40% 30% 20% 10% 0% D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 Decile Agriculture Manufacturing Services (low value added) Mining Industry Services (high value added) Source: World Bank staff calculations based on DOSM HIES/BA. Note: VA = value added. High value-added services are information and communication; Financial and insurance/takaful; Real estate; Professional, scientific and technical; Administrative and support service; Public administration and defence; compulsory social security; Education; Human health and social work; Arts, entertainment and recreation; Activities of extraterritorial organizations and bodies . Low value-added services are wholesale and retail trade; Repair of motor vehicles & motorcycles; Transportation and storage; Accommodation and food service activities; Other service; Activities of households as employers; undifferentiated goods- and services-producing activities of households for own use. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 52 By far, the highest growth in output per does not contribute much to household income worker was in high-value-added services; across the distribution (figure 4.13). Growth was richer households benefitted the most. The substantially higher in high-value-added services, growth in output per worker (figure 4.14) reflects which more than doubled over the period. Agri- both sectoral GDP growth and employment pat- culture and low-value-added services grew the terns, and is thus closely connected to growth of least, and the latter was especially affected during household income along the income distribution. the pandemic, when public health measures re- Value added per worker between 2004 and 2022 stricted contact-intense activities. This explains grew in all sectors except mining,50 which only some of the stalling in inequality reduction. employs a very small share of the population and F I G U R E 4 .1 4 OUTPUT PER WORKER GREW THE MOST IN SECTORS WHERE THE RICHEST WORK 250 Index per output worker (2004 = 100) 200 150 100 50 0 2004 2007 2010 2013 2016 2019 2022 Source: World Bank staff calculations based Agriculture Manufacturing Services (low value added) on DOSM GDP and employment by sector. Mining and Quarrying Other industry Services (high value added) F I G U R E 4 .1 5 S E C T O R O F E M P L O Y M E N T B Y E D U C AT I O N L E V E L Share of employment (%) 20 40 60 80 100 0 35 14 12 39 No Certificate ABC 24 17 9 49 Primary 7 19 7 67 Secondary 4 16 8 72 Post-secondary 1 12 5 80 Tertiary Agriculture Mining Manufacturing Construction Services Source: World Bank staff calculations based on DOSM LFS. 50. Although mining output per worker was high in 2004. 53 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA By ethnicity, while the employment incomes Most of the decline in labor income differ- of Bumiputera workers have grown rapid- ences between the Bumiputera and other ly, the absolute gap in average labor income ethnic groups has been because of an im- across ethnic groups has declined only mar- provement in endowments, especially edu- ginally. Employment incomes have risen over cation. A Oaxaca-Blinder decomposition for the time for all ethnic groups, including for Bumiput- gap in tertiary employment incomes between the era workers. The average labor incomes of the Bumiputera and other ethnic groups shows that Bumiputera grew 125 percent between 2004 and the share of the explained component, which 2022, a rate similar to that for Indian workers and constitutes the endowments, in the difference in 7 percentage points higher than the rate for Chi- average employment incomes has declined over nese workers (figure 4.16). Nonetheless, because time, whereas the share of the unexplained com- the Bumiputera started at a much lower labor ponent, the returns to those endowments, has income level, the absolute gap has not narrowed increased (figure 4.17). very much. F I G U R E 4 .1 6 R E A L M E D I A N E M P L O Y M E N T I N C O M E B Y E T H N I C I T Y, 2 0 0 4 - 2 2 1.40 6000 5000 1.30 4000 1.20 1=2004 RM 3000 1.10 2000 1.00 1000 0.90 0 2004 2007 2009 2012 2014 2016 2019 2022 Bumiputera (1=2004) Chinese (1=2004) Indian (1=2004) Bumiputera Chinese Indian Source: World Bank staff calculations from DOSM HIES/BA. F I G U R E 4 .1 7 E XPL AINED AND UNE XPL AINED DIFFERENCES IN D I S T R I B U T I O N O F T O TA L E M P L O Y M E N T EARNINGS, 2004–22 2004 2007 2009 2012 2014 2016 2019 2022 0 Percent difference in log earnings (Bumiputera - non-Bumiputera) -10 -20 -30 -40 Explained (endowments) Unexplained (returns) -50 Differences in income Source: World Bank staff calculations from DOSM HIES/BA. Note: Oaxaca-Blinder decomposition Bumiputera versus other ethnicities. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 54 The earning premium for higher education to 32 percent by 2014, 29 percent by 2019, and 27 has been declining, and educational attain- percent by 2022. The return on above secondary ment has become less important in explain- education (vocational or university) was higher ing gaps in employment outcomes. Figure 4.21 than that earned by upper secondary education, reports the average skill premium for upper sec- but also fell more markedly over the past 20 years ondary and above secondary (vocational or uni- (Figure 4.19). The return fell from 78 percent in versity) relative to lower secondary education or 2004 to 66 percent in 2014, to 61 percent in 2019, below. As educational attainment has increased and to 58 percent in 2022. For both upper sec- considerably in the country, it has become a less ondary education and above, the decline in the important source in the variation in employment education premium was more marked in the first incomes. While incomes have grown for all ed- decade (2004–14) than in the second (2014 on- ucational groups (Figure 4.18) and there is still ward), coinciding with the inequality decline and a positive and large average high-skill premium stall patterns. Workers with more than secondary (the return on higher education controlling for education, for whom the education premium has workers characteristics), the premium has de- been falling, are mainly in the services sector. clined in the past two decades. In 2004, upper Over 70 percent of workers with post-secondary secondary education would earn a return 37 per- or university education are employed in services cent higher than lower secondary; this declined (figure 4.15). F I G U R E 4 .1 8 WA G E I N C O M E G R O W T H B Y E D U C AT I O N L E V E L , 2 0 1 0 – 2 2 150 Wage income growth (2010=100) 140 130 120 110 100 90 80 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Post-secondary Primary Tertiary No formal education Secondary Source: World Bank staff calculations from DOSM HIES/BA. F I G U R E 4 .1 9 AV E R A G E S K I L L P R E M I U M , 2 0 0 4 – 2 2 100 90 Average premium of education 80 70 60 50 40 30 20 10 0 2004 2007 2010 2013 2016 2019 2022 Secondary Above secondary Source: World Bank staff calculations from DOSM HIES/BA. Note: The regressions control for gender, potential experience, education level, sector, ethnicity, strata, informality, occupation, and skill level. Since the HIES/BA only has information on the level of educational attainment, the years of education, which is needed to estimate potential experience, are proxied by the years required to complete each educational level. Alternative estimations using age and age-squared yield very similar results. 55 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Differences in the skill premium across the income distribution. However, the returns on income distribution are an important rea- higher education (vocational or university) have son income inequality persists. The returns on an upward slope, indicating that they are lower education for different quantiles are estimated for workers in poorer households, even after con- using a recentered influence function regression trolling for other individual characteristics, such of the unconditional quantile (for the 5th to the as the skill of the occupation. The estimated re- 95th percentile) of employment income on a set turn on higher education (relative to lower sec- of individual sociodemographic characteristics ondary or below) in 2022 was 46 percent at the (Firpo, Fortin, and Lemieux 2018). Figure 4.21 il- first ventil of the household income distribution lustrates the results of mapping workers’ school- and increased to 64 percent at the top ventil, in- ing premium to the rankings of their families’ per dicating that individuals in the bottom of the in- capita incomes (by ventils in the income distri- come distribution can expect the lowest returns bution).51 The return on an upper secondary (or on schooling, even though they stand to gain the lower) education, relative to below secondary most from education. education, is low but flat across the household FIGURE 4.20 S K I L L P R E M I U M B Y P E R C A P I TA H O U S E H O L D I N C O M E V E N T I L , 2 0 0 4 – 2 2 Secondary Above secondary 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 v1 v3 v5 v7 v9 v11 v13 v15 v17 v19 v1 v3 v5 v7 v9 v11 v13 v15 v17 v19 2004 2014 2019 2022 Source: World Bank staff calculations from DOSM HIES/BA. Note: The estimates are the unconditional returns on education (relative to below secondary education) in the per capita household income distribution. The regressions control for gender, potential experience, education level, sector, ethnicity, strata, informality, and occupation skill level. Since the HIES/BA only has information on the level of educational attainment, the years of education, which are needed to estimate potential experience, are proxied by the years required to complete each educational level. To link the conditional returns to workers’ positions in the (unconditional) per capita household income distribution, we follow the method in Arias et al. (2006). V = ventil. 51. The ranking of workers in the conditional earnings distribution can be taken as a proxy of their level of “ability,” or unmeasured earning determinants. We would like to link these conditional returns to workers’ positions in the (unconditional) per capita household income distribution. To do this, we follow the method in Arias et al. (2006). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 56 FIGURE 4.21 R E T U R N S O N E D U C AT I O N B Y P E R C A P I TA HOUSEHOLD INCOME VENTIL, 2004–22 1 Secondary 1 Above secondary 0.9 0.9 0.8 0.8 0.7 0.7 0.6 0.6 0.5 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0 0 20042006 2008 2010 2012 2014 2016 2018 2020 2022 2004 2006 2008 2010 2012 2014 2016 2018 2020 2022 v1 v2 v3 v4 v5 v6 v7 v8 v9 v10 v11 v12 v13 v14 v15 v16 v17 v18 v19 v20 Source: World Bank staff calculations from DOSM HIES/BA. Note: The estimates are the unconditional returns to education in the per capita household income distribution. The regressions control for gender, potential experience, education level, sector, ethnicity, strata, informality, and occupation skill level. Since the HIES/BA only has information on the level of educational attainment, the years of education, which are needed to estimate potential experience, are proxied by the years required to complete each educational level. To link the conditional returns to workers’ positions in the (unconditional) per capita household income distribution, we follow the method in Arias et al. (2010). V = ventil. The skill premium has fallen over time for all and work in lower-skilled occupations. Work- households across the income distribution. ers in the lowest decile of the household total Across all ventils of the household income dis- income distribution are more likely to be self-em- tribution, the skill premium was increasing from ployed, although the share of self-employment in 2004 to 2007, declined sharply with the 2009 the lowest decile roughly halved between 2004 global financial crisis (GFC), and then recovered and 2022 (figure 4.22). Moreover, close to 80 slightly but without returning to the upward path percent of workers in the lowest decile are infor- before the crisis (figure 4.21). The fall was more mal—which is not only related to lower wages but pronounced between 2007 and 2014 and was also lower or no benefits and protection against especially stronger for the richest households, shocks—whereas the share is closer to 20 percent which had gained the most between 2004 and for workers in the top decile (figure 4.24). Similar- 2007 (despite a less pronounced fall during the ly, 85 percent of workers in the top decile are in GFC). In contrast, the premium for secondary ed- high-skilled occupations,52 whereas in the lowest ucation was almost flat between 2004 and 2022. decile, less than 10 percent of workers are in this group, half are in routine occupations, and close Poor households earn less in the labor mar- to 20 percent remain in low-skilled occupations ket because they are more likely to be own (figure 4.23). account workers, tend to be more informal, 80 percent of workers in the lowest decile are informal—which is not only related to lower wages but also lower or no benefits and protection against shocks. 52. The classification of occupations follows Acemoglu and Autor (2011) and Bárány and Siegel (2018). High-skilled occupations include managers and managing proprietors, professionals, associate professionals, and technicians. Middle-skilled routine occupations include clerical support workers, craft and related trades workers, and plant and machine operators and assemblers. Middle-skilled nonroutine occupations include service and sales workers. Low-skilled occupations include elementary occupa- tions, which include cleaners and helpers, and laborers in mining, construction, manufacturing, and transport. 57 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.22 A C T I V I T Y I N H O U S E H O L D I N C O M E D E C I L E S 1, 5, A N D 1 0 Decile 1 Decile 5 Decile 10 100 100 100 90 90 90 80 80 80 70 70 70 60 60 60 Percentage Percentage Percentage 50 50 50 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 2004 2007 2009 2012 2014 2016 2019 2022 2004 2007 2009 2012 2014 2016 2019 2022 2004 2007 2009 2012 2014 2016 2019 2022 High skilled Medium skilled routine Medium skilled non-routine Low skilled Source: World Bank staff calculations from DOSM HIES/BA. Note: For 2004 and 2007, it is not possible to distinguish between public and private sector employees. FIGURE 4.23 O C C U PAT I O N S B Y S K I L L L E V E L I N H O U S E H O L D I N C O M E D E C I L E S 1, 5, A N D 1 0 Decile 1 Decile 5 Decile 10 100 100 100 90 90 90 80 80 80 70 70 70 60 60 60 Percentage Percentage Percentage 50 50 50 40 40 40 30 30 30 20 20 20 10 10 10 0 0 0 2004 2007 2009 2012 2014 2016 2019 2022 2004 2007 2009 2012 2014 2016 2019 2022 2004 2007 2009 2012 2014 2016 2019 2022 High skilled Medium skilled routine Medium skilled non-routine Low skilled Source: World Bank staff calculations from DOSM HIES/BA. FIGURE 4.24 100 88,6 INFORMALIT Y BY 90 Rate of informal employment, percent INCOME DECILE, 80 76,4 2009–19 70 81,3 61,8 60 53,9 60,4 46,9 50 37,6 40 46,2 32,7 30 37,2 26,1 21,4 30,7 17,6 20 27,5 23,4 20,5 18,2 18,8 10 0 1 2 3 4 5 6 7 8 9 10 Income decile Source: World Bank 2023b. 2009 2019 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 58 FIGURE 4.25 O C C U PAT I O N S B Y S K I L L A N D S TAT E WP Kuala Lumpur WP Labuan Pulau Pinang Melaka Selangor Negeri Sembilan Johor Kedah Pahang Perak Perlis Kelantan Sarawak Terengganu Sabah 0 20 40 60 80 100 Share of skills (percent) Source: World Bank staff calculations based on DOSM LFS. High skill Mid-skill Low skill The skill premium in Malaysia is declining ers’ needs. In the 2019 Enterprise Survey (World due to, among other reasons, the deficien- Bank 2020), an inadequately educated workforce cies in education quality in the country and was the joint top-third business environment ob- the inadequate creation of high-skilled jobs. stacle faced by firms, alongside electricity access, This could be a worrisome sign as regards the access to finance, and crime (figure 4.27). Overall, valuation of education by employers in Malaysia, 10 percent of the surveyed firms reported this as possibly reflecting the limitations in education a concern, which was present among 20 percent quality in Malaysia described earlier. This trans- among medium-sized firms (figure 4.27). lates into a labor supply that is not fit for employ- FIGURE 4.26 TOP THREE BUSINESS ENVIRONMENT CONSTRAINTS BY SIZE Small (5-19 employees) Medium (10-99 employees) Large (100+ employees) 30 30 30 29% 25 25 25 20 20 20 20% 18% % of Firms % of Firms % of Firms 15 17% 15 16% 15 13% 14% 10 10 12% 10 11% 5 5 5 0 0 0 ct e es y fo ely ct e rm ses y fo ely ct e cit ilit se th Bu l se f th se th at or e or its e or rk at rk at pe n tri ab rc rc al of al of xr a o d ice wo equ wo qu ec st rm es rm es rm es Ta e an ss l l in El fo ic fo ic fo ic ed d ed d at Ina at Ina in ract in ract in act e ica sin Pr lit P P Po uc uc ed ed Source: World Bank 2020. 59 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.2 7 S H A R E O F F I R M S A N D D I F F I C U LT Y T O S AT I S F Y T H E D E M A N D FOR HIGH-SKILLED L ABOR, BY FIRM SIZE Large Enterprises 34.3 49.4 13.1 Small-Medium Enterprises 32.5 49 13.3 0 20 40 60 80 100 Percentage Strongly agree Slightly agree Slightly disagree Strongly disagree Source: Business pulse survey. Another potential reason for declining re- age points over 11 years. The shares of high- and turns on education over time lies in the im- low-skilled jobs increased slightly, by 1.9 percent- balance between the expansion of education age points and 1.3 percentage points, respectively. in the country but a lack of an equivalent em- More and more people are completing secondary ployment generation. Malaysia’s transition to school, but they are staying unemployed or ob- an upper-middle-income economy was closely taining lower-skilled jobs. The unemployment linked to a rapid transition from an agricultural rate among youth under 25 years old is double the to an industrial economy (World Bank 2023a); national rate, which is over 10 percent, suggesting however, as discussed in the introduction, pro- that a lack of productive work opportunities for ductivity growth is essential to fulfil the country’s the youth is a growing concern for the country. high-income aspirations. Improving productivi- Over 70 percent of those with post-secondary ty in existing jobs through better education and and tertiary education are in the services sector, skills is part of the effort to boost productivity, as but only a third of jobs in services are high skilled is the creation of more high-skilled jobs for the (figure 4.30). Skill-related underemployment, or future. The share of employment by skill level the share of workers with tertiary education em- changed only very gradually between 2010 and ployed in semi-skilled or low-skilled jobs, is high 2021 (figure 4.28). In this period, the share of mid- specially among the youth (figure 4.29). dle-skilled jobs declined slightly, by 3.3 percent- A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 60 FIGURE 4.28 JOBS BY SKILL LEVEL , 2010–21 100 Share of employment (percent) 80 60 40 20 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Low skill Mid-skill High skill Source: World Bank staff calculations based on DOSM LFS. FIGURE 4.29 S K I L L- R E L AT E D U N D E R E M P L O Y M E N T B Y A G E , 2 0 1 0 – 2 1 80 Underemployment rate (percent) 60 40 20 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 15 – 24 25 – 34 35 – 44 45 – 54 55 – 64 Source: World Bank staff calculations based on DOSM LFS. Thus, Malaysia faces a dual problem of declin- likely conceals increasing wages and skill premi- ing higher education premiums but a short- ums for skilled workers in selected industries. age of skills. The unmet skill demand reported by firms should lead to a premium in wages for High-skilled jobs are particularly missing in scarce skilled workers. But the premium paid to the lagging states, especially for the Bumiput- those with higher education is falling. How is this era in East Malaysia, but more high-skilled reconciled? First, if skilled jobs are created at a jobs are also needed in services and across rate slower than tertiary graduation rates, then all ethnic groups in the Peninsula. The share some are compelled to take lower skilled jobs, of high-skilled jobs is the highest in mining, while as indicated by the very high rate of skill-related smaller shares are also in manufacturing, con- underemployment among the youth. This drives struction, and services. Ten percent of services down the national average return on higher ed- employment entails low-skilled occupations, 58 ucation and leads to a falling higher education percent entails middle skilled occupations, and premium, as is seen. Falling premiums and the the remaining 34 percent entails high-skilled high cost of higher education could become a occupations. By location, half of the workers in disincentive for future workers to invest in great- Kuala Lumpur are in high-skilled occupations, er human capital. At the same time, education whereas the share is close to 20 percent in the is decreasingly a proxy for skills, as it is often in lagging states (figure 4.25). Across ethnicity and lower-income countries. The skills many students location, the share of employment in high-skilled acquire in tertiary education are not those need- jobs is the lowest among the Bumiputera in East ed by higher-value-added firms. Thus, the falling Malaysia, followed by the Indian (who are primar- education premium is a national average, which ily in the Peninsula) (figure 4.31). 61 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.30 SKILLS BY SECTOR 100 10% 9% 9% 24% Share of employment (%) 80 36% 41% 60 58% 64% 48% 40 55% 49% 20 27% 28% 34% 8% 0 g es re in n g ic tio rin tu in rv M ul uc tu Se ric ac tr Ag ns uf Co an M Low skill Mid-skill High skill Source: World Bank staff calculations based on DOSM LFS. FIGURE 4.31 SKILLS BY E THNICIT Y Peninsular East Malaysia 100 5% 5% 8% 11% 10% Share of employment (%) 80 55% 60% 60 60% 56% 67% 40 20 41% 36% 32% 32% 23% 0 a r an a e la er er es di su ut ut in In in Ch ip ip en m m Bu Bu eP es in Ch Low skill Mid-skill High skill Source: World Bank staff calculations based on DOSM LFS. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 62 B O X 4 .1 G E N D E R G A P S I N T H E M A L AY S I A N L A B O R M A R K E T Women in Malaysia are still less likely than men to participate in the labor market. Labor force partici- pation is high in Malaysia, at close to 70 percent, but while women have been increasingly joining the labor force, there is still a gap of about 25 percentage points in the male versus female labor force participation rate in the country. That is one of the highest gaps in the East Asia and Pacific region. Women with lower education are less likely to participate in the labor market. Malaysian women and men have very similar educational attainment rates, but the average labor market participation rate among women conceals the substantial variation by education level. Close to 90 percent of women with tertiary education, but less than half of the women with primary or lower education, participate in the labor market. F I G U R E B 4 .1 .1 F I G U R E B 4 .1. 2 E D U C AT I O N A L AT TA I N M E N T F O R M E N A N D MALE AND FEMALE L ABOR FORCE WOMEN, 2022 PA R T I C I PAT I O N ( B Y E D U C AT I O N ), 2 0 1 0 – 2 1 100% 100 80% 80 60 LFPR (%) 60% 40% 40 20% 20 0% 0 Men Women 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 University and above Vocational and diploma Women - Tertiary Men Upper Secondary Lower secondary or less Women - Post-secondary Women - Secondary Women - No certificate Women - Primary Source: World Bank staff calculations based on Source: World Bank staff calculations based on DOSM LFS. DOSM HIES/BA. Note: LFPR = labor force participation rate. Participation rates among women vary significantly as they pass through their life cycle stages; getting married and having children is associated with lower probability of women participating in the labor market, but not for men. A look at the labor market participation rates for men and women who are at different life-cycle stages yields a decent amount of information on how these key life events affect their participation in paid employment. Recent work by Kleven et al. (2023) for 134 countries shows that a woman’s probability of being employed declines during the 10 years after the birth of her first child (“motherhood penalty”). Malaysia, where the above probability is lower by 45 points, ranks eighth in terms of how much having a child lowers the probability of employment. In contrast, there is no evidence of a “fatherhood penalty” in the country. Other countries in East Asia and Pacific, such as the Philippines (25.5), Indonesia (20.5), Thailand (5.1), and Vietnam (1.2), have much smaller motherhood penalties than Malaysia. For some countries, especially lower-income countries, the penalty is appar- ent even before childbirth when a woman marries, since marriage is often closely followed by children. The motherhood penalty also relates to the care responsibilities in the household, especially respon- sibilities associated with young children. The analysis of Kleven et al. (2023) based on the birth of the first child shows the motherhood penalty as marriage and fertility decisions are made. Further, a look at the age of the youngest child shows when a mother could expect to return to work if school is the main care service and young children require more care—a task that typically falls on women. Using the Household Income and Expenditure Survey data but looking at the age of the youngest child, we show that a similar pattern is found in Malaysia. Men who were never married have participation rates of about 50 percent, regardless of their education level. Rates are similar among women with more than secondary education who were never married. When they marry, but before they have children, participation rates increase, although the increase is stronger for men. But then, participation rates decline by 6 percent for highly educated women with young children, whereas they continue increasing for men with young children, reaching over 95 percent. The participation rates for highly educated women recover once children grow up, reaching about 70 percent. Participation also declines for women with secondary or lower education, although this occurs before they have children. 63 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E B 4 .1 . 3 F I G U R E B 4 .1.4 W O R K PA R T I C I PAT I O N A C R O S S T H E W O R K PA R T I C I PAT I O N A C R O S S T H E LIFECYCLE FOR MEN AND WOMEN (AGE OF LIFECYCLE FOR MEN AND WOMEN (AGE OF THE YOUNGEST CHILD) THE OLDEST CHILD) 2019 First Child 100 0.5 80 Employment Impact 0.0 60 40 -0.5 20 Child Penalty = 0.453 (0.019) -1.0 0 1 2 3 4 5 6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 10 Event Time (Years) Series 1 Series 2 Men Women Series 3 Series 4 Source: World Bank staff calculations based on the DOSM Source: Kleven, Landais, and Leite-Mariante 2023. HIES/BA. Women who do work earn less than men. Women who are working earn less than men, even though women are less likely to be informally employed (31 percent men vs. 22 percent of women), more likely to be in mid- dle-skilled nonroutine and high-skilled occupations (26 percent vs. 30 percent), and more likely to be in services, especially high-value-added services (27 vs. 44 percent). The gap has been closing over time, but in 2022, women still earned about 20 percent less than men on average. This results in a declining but positive gap between men and women when decomposing the wage difference while controlling for a set of sociodemographic characteris- tics of women (age, education, sector of employment, location, and marital status). Sociodemographic characteris- tics of women favor their wages, while unexplained factors—the returns to those endowments—work against them. Lower earnings for women result from women working fewer hours. However, women work slightly fewer hours than men, which means lower total earnings, even if their per hour wage is higher than men’s. In 2019, before the COVID-19 pandemic, women earned a 4 percent higher wage than men for each hour worked. But they also worked slightly fewer hours than men—over half a day less per month in 2019; this results in 3 percent lower total earnings on average, even when women earn a higher per hour wage. Performing the same wage decomposition but using hourly wages instead of total wages shows a negative gap after 2017, which means women are outearning men. The Salaries and Wages Survey (SWS) data used to perform these decompositions do not have informa- tion on whether men and women have children, but when performing the same decomposition for the subset of married people, the negative gap in hourly wages is still there, but is much smaller, whereas the gap in total wages is quite large and positive. This is consistent with the previous evidence with respect to the lifecycle impacts on women’s work and a vast literature about the long-lasting impacts of childbirth-related work interruptions on skills and human capital accumulation, missing out on (improved) job opportunities, and women taking up more flexible or part-time jobs, which let them better balance their careers with care responsibilities (e.g., Goldin 2021, Kleven 2024). Malaysia is losing out on this human capital potential. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 64 F I G U R E B 4 .1 .5 F I G U R E B 4 .1. 6 S K I L L L E V E L AT W O R K B Y G E N D E R , 2 0 2 2 SHARE OF EMPLOYMENT BY SECTOR AND GENDER, 2022 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Men Women Men Women Service (high) Services (low) High skilled Middle-skilled routine Industry Manufacturing Middle-skilled Low-skilled non-routine Mining Agriculture Source: World Bank staff calculations based on Source: World Bank staff calculations based on the based on DOSM HIES/BA. DOSM HIES/BA. F I G U R E B 4 .1 . 7 F I G U R E B 4 .1. 8 O A X A C A- B L I N D E R D E C O M P O S I T I O N F O R M E N O A X A C A- B L I N D E R D E C O M P O S I T I O N F O R V S . W O M E N . D I F F E R E N C E S I N T O TA L A N D MARRIED MEN VS. MARRIED WOMEN. H O U R LY E A R N I N G S , 2 0 1 0 – 2 2 D I F F E R E N C E S I N T O TA L A N D H O U R LY EARNINGS, 2010–22 Difference in log earnings (Male - Female) % Difference in log earnings (Male - Female) % 10 45 8 40 35 6 30 4 25 2 20 0 15 -2 10 5 -4 0 -6 -5 -8 -10 1 2 3 4 5 6 7 8 9 10 11 12 20 0 20 1 20 2 20 3 20 4 15 20 6 20 7 18 20 9 20 21 1 1 1 1 1 1 1 1 20 20 20 20 Difference (M-F) total Difference (M-F) hourly Difference (M-F) total Difference (M-F) hourly Source: World Bank staff calculations based on DOSM Salaries Source: World Bank staff calculations based on DOSM Salaries and Wages Survey. and Wages Survey. Note: The data in the SWS focus on wages in the private Note: The data in the SWS focus on wages in the private sector; the informal sector and/or self-employed workers are sector; the informal sector and/or self-employed workers are thus not included. thus not included. The decompositions control for age, age-squared, educa- The decompositions control for age, age-squared, education tion-level dummies, sector dummies, ethnicity dummies, strata level dummies, sector dummies, ethnicity dummies, strata dummies, and married dummy. dummies, and married dummy. 65 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Hard work, although recognized by Malay- Negeri Sembilan and Kelantan (figure 4.33). This sians as an important determinant of up- perception could partly reflect the falling premi- ward mobility, is not perceived as currently ums for higher education and limited high-skilled leading to better economic prospects. Most job opportunities, which could subsequently lead respondents in the perceptions survey view hard to lower efforts, for example, underinvestment in work (68 percent) and high education (22 per- skill upgrade or entrepreneurial activity, as well as cent) as the most important factors for upward to lower aspirations, especially among the youth. mobility. This view is somewhat the same regard- Why many have a pessimistic view of the rewards less of a respondent’s income class, either actual on hard work, but most believe their children will or perceived. While about half of the respondents do as well as or better than them, deserves fur- in the perceptions survey think that it is easy for ther research. For example, this could reflect the people to improve their economic prospects if optimism that parents often feel about their own they work hard, another 40 percent think that it children,53 or an unwillingness to report beliefs is difficult or impossible (figure 4.32). This view is about their prospects that are less optimistic but prominent among the Indians and those living in which they truly hold. FIGURE 4.32 I M P O R TA N C E O F H A R D W O R K F O R E C O N O M I C P R O S P E C T S It is easy for people to improve their 47% 53% economic situation if they are willing to work hard 49% People find it difficult to improve their 41% 30% economic situation even if they work hard 35% It is impossible for people to improve their 6% 6% economic situation even if they work hard 8% 4% Don't know/don't want to answer 7% 6% Indian It is easy for people to improve their 2% Chinese 3% economic situation even if they don't work hard 2% Bumiputera Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: In your opinion, which of the following four statements best describes the current reality in Malaysia? FIGURE 4.33 P E O P L E F I N D I T D I F F I C U LT T O I M P R O V E T H E I R E C O N O M I C S I T U AT I O N E V E N I F T H E Y W O R K E D H A R D Negeri Sembilan 62% Kelantan 49% FT Putrajaya 46% FT Labuan 41% Selangor 40% Penang 39% FT Kuala Lumpur 37% Perak 36% Sabah 35% Sarawak 30% Pahang 30% Kedah 25% Johor 23% Terengganu 22% Melaka 10% Perlis 1% Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: People find it difficult to improve their economic situation even if they worked hard. 53. For example, Sharot (2011). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 66 INEQUALIT Y FRAMEWORK: SHOCKS While poverty in Malaysia has declined sharp- The contribution of household-specific risk ly over time and is currently quite low, many to risk-induced vulnerability is almost thrice more people are still considered vulnerable as much as the contribution of communi- to falling into poverty. The income-generating ty-level risk. Further, risk-induced vulnerability process is affected by shocks, but different house- can be further divided into that originating from holds are more likely to be affected by them. Us- shocks affecting an entire community (covariate ing the cross-sectional variation in consumption, risk) or only a specific household (idiosyncratic we develop estimates of vulnerability to poverty risk). In Malaysia, the contribution of house- and economic security (see box 3.1). The extent of hold-specific risk to risk-induced vulnerability vulnerability to poverty depends on the risk ex- is almost thrice as much as the contribution of posure and the ability of households to cope after community-level risk. It is not uncommon for this the occurrence of a shock, as well as access to to be the case; in five African countries, the ratio safety nets to potentially limit the shock’s impact of idiosyncratic to covariate risk is about 4 (Sk- on the current welfare. While poverty has de- oufias and Baez 2021) and 4.8 in the Philippines clined substantially in Malaysia,54 the share of the (World Bank 2022e), while in Viet Nam (World vulnerable population—or those with more than Bank 2022f), a country with high exposure to cli- a 20 percent chance of falling into poverty—has mate shocks, it is 1.2. fallen much less: from 15 percent to 11 percent of the population from 2004 to 2022 (figure 4.34). Within the country, vulnerability to pover- ty was highest in rural areas and in the East In Malaysia, most of the vulnerable were in region, and covariate risks had the highest this group because their incomes are high- contribution to risk-induced vulnerability ly volatile to shocks. Vulnerability can either for the households in these places. In 2019, vul- arise due to chronic insecurity, or it can be risk nerability was twice as high in rural (24 percent of induced. If a household’s expected income is be- the population) than in urban areas (12 percent) low the poverty line, it is classified as chronically (figure 4.35). In the East region, vulnerability was insecure. If the household expected income is also substantially higher than the national aver- above the poverty line but may fall below it in age, at 22 percent, and for the Bumiputera, the some states of the world due to shocks, then this rate of vulnerability was 18 percent. Of all the is called risk-induced vulnerability. In 2019, before groups, the contribution of idiosyncratic risk to the pandemic, 90 percent of the vulnerable in risk-induced vulnerability was slightly lower than Malaysia were so due to risk-induced vulnerability. that in the country as a whole only in rural areas and the East region (2.6 and 2.5 times more, re- spectively), although the share of chronic insecu- rity was the highest in these places. FIGURE 4.34 100 7% 8% 10% 13% 16% VULNERABILIT Y 90 13% 22% 27% 29% 15% 17% OVER TIME IN 80 21% M A L AY S I A , 70 15% 17% 25% 2004–22 17% 28% 60 15% 20% 29% 31% 50 15% 15% 22% 40 15% 21% 20% 30 16% 19% 50% 44% 14% 20 41% 31% 15% 11% 10 22% 16% 10% 10% 0 2004 2007 2009 2012 2014 2016 2019 2022 Source: World Bank staff calculations Upper class Economically secure Aspiring Vulnerable Poor based on DOSM HIES/BA. Economic security with P(20%) 54. Note that these poverty figures do not correspond to the official poverty rates. Instead, as in Rongen and Lanjouw (2024), they use an absolute threshold for poverty that is kept constant over time. 67 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 4.35 V U L N E R A B I L I T Y B Y G R O U P, 2 0 1 9 40 3.5 Vulnerabilty (percentage) 35 3.0 30 2.5 25 2.0 I/C risk 20 1.5 15 1.0 10 5 0.5 0 0.0 l n l a st a a na ra ul er er ba Ea Ru ns tio ut ut Ur ni ip ip Na Pe m m Bu Bu n- No Idiosyncratic / covariate (right axis) Chronic insecurity Risk-induced vulnerability Source: World Bank staff calculations based on DOSM HIES/BA. Vulnerability also varies substantially across percent), East Malaysia states (19 and 24 percent states; in states where vulnerability was the in Sabah and Sarawak, respectively), and Tereng- highest, so was the contribution of covariate ganu (20 percent) (figure 4.36). Those were also shocks. Less than 3 percent of the population in the states where covariate risks mattered rela- Kuala Lumpur was considered vulnerable. Vul- tively more. Malaka and Perak stood out as places nerability rates were lower than 10 percent also with average vulnerability levels but where co- in Putrajaya and Labuan. In contrast, the rates variate risks had relatively greater importance. of vulnerability were the highest in Kelantan (29 FIGURE 4.36 V U L N E R A B I L I T Y B Y S TAT E , 2 0 1 9 40 7.0 Vulnerabilty (percentage) 35 6.0 30 5.0 25 4.0 I/C risk 20 3.0 15 10 2.0 5 1.0 0 0.0 r h an a n ng ng k s or nu h ak r n ya pu ho rli ra ak la la ba ua ng ja w nt ha na ga Pe bi Ke Pe m Jo el ab Sa ra ra la la m Pi Pa ng M Lu ut Sa .L Ke Se Se u re .P la .P la Te ri ua W .P Pu ge W .K Ne .P W Idiosyncratic / covariate (right axis) Chronic insecurity Risk-induced vulnerability Source: World Bank staff calculations based on DOSM HIES/BA. COVID-19 is an example of a covariate shock slower among poor and vulnerable groups. As from which Malaysia is still recovering. But of May 2022, households with lower incomes, even a covariate shock like the pandemic young workers, individuals with low education, can affect some people disproportional- and business owners and informally employed ly more. Despite an impressive economic re- workers tend to continue experiencing econom- bound across the board in 2022, the World Bank’s ic shocks. Work stoppages among prepandemic High-Frequency Phone Survey (HiFy) Round 355 lower-income workers stood at 25 percent com- in Malaysia found that recovery continued to be pared with 15 percent among their higher-income 55. The HiFy Survey Round 3 was conducted in April/May 2022. The survey involved 25- to 30-minute telephone interviews of over 1,000 panel respondents, who continuously participated in three survey rounds (Round 1 was conducted in May/June 2021, and Round 2 in October/November 2021). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 68 peers. Moreover, nearly 70 percent of lower-in- children from lower-income households, who come households self-assessed themselves as were less likely than the children of more affluent having inadequate financial resources to meet households to engage in online or mobile learn- their monthly basic needs, and more than 60 ing activities based on distance learning, despite percent of these households reported having no several government-sponsored educational assis- savings. The slower return of poor and vulnerable tance programs. Moreover, even before the pan- households to work, and the depletion of their demic, significant learning gaps existed between financial resources, makes their recovery more students from socioeconomically advantaged challenging. Uneven recovery is also evident and disadvantaged households in Malaysia; these across regions; households in the East Malaysia gaps widened over time.56 region experience slower progress in regaining employment than households in other regions. Shocks also affect mobility, but different Slower progress in the East Malaysia region might shocks have different impacts across the in- not be surprising. In 2022, the absolute poverty come distribution. Because shocks affect the rates in Sabah and Sarawak of East Malaysia accumulation of assets for the future, they impact were among the highest in the country—3.0 and the ability of poorer households to move across 1.7 times the national poverty rate, respectively. the income distribution. Figure 4.37 shows the Meanwhile, urban poverty rose from 3.8 percent percentage of people in Q1 (Q5) who stayed in Q1 in 2019 to 4.5 percent in 2022, partly due to great- (Q5) and the percentage of people who moved up er employment disruptions in the services sector (down). Mobility estimates for Q1 barely changed during the pandemic, whereas rural poverty fell after the GFC, even though Malaysia was severely from 12.4 percent to 12.0 percent in the same pe- hit by this shock. This does not mean that poor riod. households were not affected, but that they were less affected relative to richer households, perhaps Besides financial losses, the pandemic’s non- because the GFC severely hit financial assets, monetary impacts may eventually be costlier which were disproportionately held by the rich. in the longer term. The period during which In fact, when looking at mobility estimates for the schools remained closed due to pandemic re- top quintile, the probability of staying safely there strictions has brought to the fore issues surround- fell by 2.5 points after the GFC. This pattern is not ing the risks of heightened learning losses among observed during the COVID-19 pandemic shock. students across the world. For Malaysian stu- The share of individuals who were “trapped” at dents, pandemic-related school closures are esti- the bottom or “secure” at the top barely changed, mated to lead to additional learning losses of up perhaps reflecting the broad nature of this shock to 1.3 years, which could result in future average affecting everyone across the income distribu- annual earning losses of up to US$2,320 per stu- tion. dent. The impact is expected to be worse among FIGURE 4.37 S H O C K ’ S I M PA C T S O N M O B I L I T Y I N Q 1 A N D Q 5 GFC impacts on Q1 and Q5 mobility 65 Quintile 1 Quintile 5 60 55 50 Percent 45 40 35 30 GFC Covid-19 GFC Covid-19 Trapped (pre-shock) Escape (pre-shock) Fall (pre-shock) Secure (pre-shock) Trapped (post-shock) Escape (post-shock) Fall (post-shock) Secure (post-shock) Source: World Bank staff calculations based on Rongen and Lanjouw (2024). Note: The estimates are the percentage of people in Q1 (trapped) and Q5 (secure) staying in same quintiles. GFC pre-shock spans 2004–09 and post-shock spans 2004–12. COVID-19 pre-shock spans 2014–19 and post-shock spans 2014–22. 56. According to PISA 2009, socioeconomically advantaged students in Malaysia outperformed their disadvantaged peers in reading by 53 points. By 2018, this gap had widened to 89 points, meaning that the performance of advantaged students improved where- as that of disadvantaged students declined over this period (World Bank 2021b). 69 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA The risks associated with climate hazards are Barbier 2018). For instance, poorer households another source of covariate shocks. Globally, are more likely to reside in flood-prone regions, Malaysia ranks 132 out of 168 countries in terms where insurance and adaptive or preventive mea- of the population exposed to climate shocks; 23 sures against high water are poorer. percent of the population is exposed (Doan et al. 2023).57 Floods have historically been the main Most districts under high climate hazard risk hazard in Malaysia; they occur particularly during are relatively wealthier, but pockets of rela- monsoon, when heavy rainfall can lead to river tively elevated risk and low living standards overflow and flash floods, impacting both urban exist. The distributional impact of climate risks and rural areas. The coastal and riverine plains can be examined by overlaying hard risk (expect- are flood-prone areas; they have become increas- ed annual impact [EAI] or expected annual ex- ingly populated, and in the past decades, signif- posure [EAE]58) with relative income; it shows icant river and pluvial floods have been record- the locations where risk is most likely to translate ed almost every year (Shah et al. 2017). Some of into severe impacts on the poorest and vulnera- the most intense severe events include the 2004 ble households. In the absence of small-area esti- and 2014 floods, which impacted the eastern mates of poverty or income, here we use the Rel- states of peninsular Malaysia, namely, Kelantan, ative Wealth Index (RWI) score of Blumenstock, Terengganu, and Pahang. Coastal floods are also Cadamuro, and On (2015) and Chi et al. (2022). an important hazard; it is estimated that about This crossing provides ranks in a 3 x 3 matrix; the 60–70 percent of the total population lives with- result is nine possible scores, which range from in 5 kilometers (km) of the coastal perimeter. The low risk/high living standards to high risk/low worst coastal impacts on population occur along living standards. River floods pose mostly low the coastal perimeter in the states of Sabah, Se- and mild risk across Malaysian districts; the only langor, and Sarawak. The largest absolute and rel- district under high river flood risk is Sibu (Sar- ative built-up area impact is in Langkawi district awak), which is classified as a medium RWI dis- (Kedah state), where about 30 percent of built-up trict. Two additional districts in Sarawak (Kapit area could be lost due to floods. In terms of heat and Song) are also of concern due to moderate stress, Wet-Bulb Globe Temperature (WBGT) flood risk and low relative living standards. For values above 32°C are common in most coastal costal floods, the pattern is different; most high- states, including the metropolitan areas of Kuala risk districts have higher living standards, which Lumpur and other important urban centers. are reflected in a higher RWI score. A couple of districts with medium living standards and high The risks and their impacts are however dis- risk are in Sandakan and Semporna in Sabah and tributed unevenly across the country. The Sebak Bernam in the Peninsula. For landslides, risk of economic impacts due to shocks depends all high-risk districts have relatively high living on the probability of a hazard occurring and the standards, meaning that households are likely to intensity with which it occurs, the exposure to be able to implement risk reduction measures at peoples and their assets, and the vulnerability least partially. The three districts where landslide of peoples’ incomes and livelihoods. In general, risk is higher and living standards are medium are poorer households are the least resilient to cli- in Sabah (Penampang, Sandakan, and Tuaran). mate impacts because they tend to reside in more Finally, there is a noticeable inverse relationship disadvantaged and hazard-prone areas, and have between heat stress risk and poverty because less access to critical services like health, educa- low-risk areas tend to be less prosperous moun- tion, and early-warning systems; their assets are tainous areas (figure 4.38).59 more vulnerable to hazards (Hallegatte, Fay, and 57. Further, from an international standpoint, due to the country’s relative low poverty rate, high income, and high coverage of basic services, it has a low proportion of the exposed population, which is highly vulnerable to suffering severe economic impacts due to shocks. 58. The EAI quantifies the distribution of risk across a country according to geospatial risk modeling and past observations. When probabilistic hazard scenarios are not available to calculate the EAI, the EAE is calculated for a selected hazard threshold. See appendix V for methodology. 59. This pattern is likely to be even stronger than what is shown because this hazard model does not account for the heat island effect, which could increase the intensity of extreme temperature events in (relatively wealthier) cities. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 70 FIGURE 4.38 C L I M AT E H A Z A R D R I S K A N D R E L AT I V E L I V I N G S TA N D A R D S AT T H E D I S T R I C T L E V E L A. River floods B. Coastal floods Risk (EAI/KM2) Risk (EAI/KM2) RWI RWI C. Landslides D. Heat stress Risk (EAE/KM2) Risk (EAE/KM2) RWI RWI High risk Mid risk Low risk Source: World Bank staff calculations based on CRED/UCLouvain (2023), EM-DAT (www.emdat.be), and http://www. povertymaps.net/. Note: Cut-off values to define risk level are country and hazard specific. Medium-risk ranges are in EAI/EAE/km2: river flood [0.1–1], coastal flood [0.01–0.05), landslide [25–250), heat stress [100–300). The metrics for EAE/EAI are computed relative to the ADM area (km2) in order to normalize the distribution of values; this is to avoid outliers that would cause an unrepresentative quantile splitting. RWI cut-off values for the range for medium living standards are [-0.25–0.25). The RWI values are weighted using the 2020 population. EAE = expected annual exposure; EAI = expected annual impact; km2 = square kilometers; RWI = Relative Wealth Index. 71 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA The incidence and impact of climate-related built-up area is expected to increase more prom- shocks will be increasingly important in the inently, due to the expected sea level rise, by 40 future. Climate hazards will be an increasing and 29 percent of the baseline, respectively, by source of risk in Malaysia. The impact of river 2050. Regarding heat stress (figure 4.39), more floods on population and built-up area is expect- frequent moderate (WBGT > 23°C) as well as ed to increase by, respectively, 17 and 12 percent extreme (WBGT > 30°C) heat stress events can of the baseline by 2050 (table 4.2).60 It is estimat- potentially translate into unprecedented adverse ed that by 2030, floods can cost Malaysia about 4 health impacts and mortality; the coastal plains, percent of the total output and almost 330,000 which are the most populated areas, will be the jobs by 2030 (World Bank and BNM 2024). The most affected by this increase. impact of coastal floods on the population and TA B L E 4 . 2 E X P E C T E D A N N U A L I M PA C T F R O M A P R O B A B I L I S T I C A N A LY S I S O F F L O O D R I S K T O P O P U L AT I O N A N D B U I LT- U P A R E A   River flood Costal flood   2050 (change from baseline) 2050 (change from baseline)   SSP3 7.0 SSP3 7.0 Return Impact on Impact on Impact on Impact on period population built-up area population built-up area (years) (%) (%) (%) (%) RP 5 20 18 50 30 RP 10 15 13 50 29 RP 20 8 12 0 28 RP 50 7 12 33 27 RP 100 8 11 33 26 RP 200 7 10 33 26 RP 500 8 8 67 26 RP 1,000 6 7 25 26 Annual 17 12 40 29 Source: World Bank staff calculations based CRED/UCLouvain (2023), EM-DAT (www.emdat.be), and Coupled Model Intercompari- son Projects (CMIP6). Baseline in 2020. Note: RP = return period; SSP = Shared Socioeconomic Pathway. FIGURE 4.39 D AY S W I T H WA R M N I G H T S E X C E E D I N G 3 5 ° C , 1 9 5 0 – 2 1 0 0 Baseline 250.0 200.0 150.0 100.0 50.0 0.0 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 2060 2065 2070 2075 2080 2085 2090 2095 2100 Historical SSP1 - 2.6 SSP2 - 4.5 SSP3 - 7.0 Source: World Bank staff calculations based on CRED/UCLouvain (2023), EM-DAT (www.emdat.be), and Coupled Model Intercom- parison Projects (CMIP6). 60. See appendix V for methodology. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 72 INEQUALIT Y FRAMEWORK: TA X E S A ND PUBL IC SPE NDING The Malaysian system of taxes and transfers In cash terms (excluding health and education), is progressive and reduces inequality, but it the richest 10 percent are the main contributors trails many UMIC peers and even some low- to the fiscal system; they contribute 63 percent er-middle-income countries (LMICs) in the of the total. Another 15 percent comes from the inequality reduction that it achieves. Taxes second-richest decile. While taxes, transfers, and and transfers can help mitigate shocks and sup- subsidies increase the disposable income of the port households through the income-generating poor, the shares of benefits received by poor- process. Taxes, cash transfers, and subsidies in er and higher-income households are relatively Malaysia reduce inequality by 2.4 points from the similar. The majority of the inequality reduction pre-fiscal level, which is close to the UMIC aver- through the fiscal system results from health and age. When noncash health and education benefits education benefits. When these in-kind benefits are included, Malaysia’s inequality reduction is 4.1 are included, all but the top two deciles are the points higher, for a total of 6.5 points (figure 4.40). net beneficiaries of the fiscal system (figure 4.41). While the fiscal system achieves some inequality Nonetheless, the contribution of education and reduction and is progressive, it trails many UMIC health spending to the fiscal system in Malaysia peers. It ranks 18th out of 25 UMICs—10 LMICs is lower than the average for UMICs. In addition, perform better than Malaysia (figure 4.42). while the benefits due to education and health are important for mobility and long-term pros- However, while the cash impacts are near the perity, they are not cash benefits and thus do not UMIC average, the contribution of education help meet households cover their daily living ex- and health spending is considerably lower. penses. F I G U R E 4 .4 0 G I N I I N D E X B E F O R E A N D A F T E R F I S C A L P O L I C Y, 2 0 1 9 ( P E R C E N TA G E ) 50 43.1 40.5 40 40.5 Points 30 36.8 20 Market income After direct taxes After indirect taxes After health plus direct transfers and subsidies and education Source: World Bank staff calculations based on 2019 HIES/BA. F I G U R E 4 .4 1 TA X E S A N D T R A N S F E R S B Y I N C O M E D E C I L E , 2 0 1 9 ( P E R C E N TA G E O F M A R K E T I N C O M E ) 60 Health 40 Education received Benefits Indirect subsidies Percent Cash transfers 20 Direct taxes Indirect taxes 0 Net cash impact Taxes paid Net fiscal impact -20 1 2 3 4 5 6 7 8 9 10 Household market income per capita decile Source: World Bank staff calculations based on 2019 HIES/BA. 73 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Change Change In Gini Index In Gini Index (points) (points) -25 -20 -25 -15 -20 -10 -15 -5 -10 0 -5 0 F I G U R E 4 .4 Spain Spain Uruguay Uruguay High Panama Panama United States United States Note: H+E = health Cash Croatia Croatia Mauritius Mauritius and High income 2 income Romania Romania taxes South Africa South Africa Argentina Argentina Brazil Brazil Mexico Mexico education. Namibia Namibia Georgia Georgia Venezuela. RB Venezuela. RB and transfers Cash taxes and transfers Costa Rica Costa Rica Bostwana Bostwana Dominican Republic Dominican Republic Colombia Colombia China China Equador Equador C H A N G EUpper-middle Thailand Thailand IN GINI IN Peru Peru Iran Iran Turkiye Turkiye Upper-middle income income Belarus Belarus Malaysia Malaysia Jordan Jordan Malaysia Malaysia Russian Fed Russian Fed Albania Albania Guatemala Guatemala A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Paraguay Paraguay Indonesia Indonesia Eswatini Eswatini Lesotho Lesotho Source: World Bank staff calculations based on 2019 HIES/BA. World Bank 2022c. Zambia Zambia Tunisia Tunisia Kenya Kenya NT Ukraine Ukraine Honduras Honduras F I S C A L I N S T R U M E N T S ( P E R C E N TA G E P O I N T S ) El Salvador El Salvador Mongolia Mongolia In-kind spending on health and education In-kind spending on health and education Bolivia Bolivia D E X D U E T O D I F F E R ELower-middle India India Nicaragua Nicaragua Moldova Moldova Egypt Egypt Lower-middle income income Tanzania Tanzania Ghana Ghana Sri Lanka Sri Lanka Comoros Comoros Côte d’Ivoire Côte d’Ivoire Uganda Uganda Burkina Faso Burkina Faso Togo Togo Net fiscal impact Net fiscal impact Mali Mali Ethiopia Ethiopia Niger Niger Gambia. The Gambia. The Low income Low income Tajikistan Tajikistan Guinea Guinea 74 05 ADDRESSING INEQUALITY AND PROMOTING MOBILITY: WHAT CAN BE DONE? 75 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA I nequality is a high priority Public opinion on whether inequality is ac- for the Government of Ma- ceptable is divided. More than half of the re- laysia. Inequality has been spondents of the perceptions survey think that high on Malaysia’s policy agen- an income gap could be acceptable, but mostly da since the New Econom- (3 in 10 respondents) if the basic standard of liv- ic Policy of 1970, which saw ing is maintained and the cost of living is afford- group-based ethnic disparities able. Malaysian’s tolerance on inequality differs by as the main obstacle to address for achieving economic status: acceptance of inequality, by the national unity and poverty reduction. While the actual or the perceived income class, increases policy emphasis has drifted in subsequent five- with income level. Nonetheless, the gradient of year national development plans, inequality has the acceptance pattern is much steeper between remained prominent. The country’s most recent and within groups according to respondents’ per- key strategic vision document, the MADANI ceived income. As figure 5.1 shows, acceptance Economic Framework, elevates inclusive growth level differed by up to 24 percentage points be- and equality of opportunities as key missions for tween the poorest and richest groups based on the country. the perceived income; the difference was three times as much as the difference when the groups were ranked by their actual income. F I G U R E 5 .1 A C C E P TA N C E O F I N C O M E G A P B Y A C T U A L A N D PERCEIVED INCOME CL ASS By Actual Income Class By Perceived Income Class 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 In certain circumstances. the income gap is common and acceptable In any case. the income gap should not occur and cannot be accepted at all Source: World Bank-DOSM Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the following statements, which one do you agree with more? (Under certain circumstances, the income gap is com- mon and acceptable/In any case, the income gap should not occur and cannot be accepted at all/Don’t know/Don’t want to answer.) Yet, most Malaysians, rich and poor, think in- ture (within the next three years) (figure 5.2). This equality in the country needs to be addressed holds true for the poorest as well as the richest urgently. When asked about the current inequal- quintile (70 percent), although the level of ur- ity in the country, about three in five Malaysians gency declines as people become wealthier. This think the government should address it either final section looks at how this might be achieved. immediately (less than a year) or in the near fu- A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 76 F I G U R E 5. 2 P R E F E R E N C E F O R T H E T I M E TA K E N F O R T H E M A L AY S I A N G O V E R N M E N T T O A D D R E S S I N E Q U A L I T Y I N M A L AY S I A , B Y P E R C E I V E D I N C O M E C L A S S Don't know/don't want to answer Not at all Not now (Over 5 years) In the long term (More than 3 years – 5 years) In the near future (1 – 3 years Immediately (Less than 1 year) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the following question: In your opinion, how long do you want the Malaysian government to take to address the issue of inequality in the country? The report concludes by outlining key policy strengthening the social protection delivery priorities for restarting inequality reduction chain to better support poorer households and in Malaysia. Many policies that have worked in make better use of existing budgets; (3) a series other countries, some of which have been tried of fiscal reforms are discussed, which illustrate in Malaysia as well, are not new. The report brings how these policy priorities can be financed, and nuances to improve how these policies work in the political economy of these reforms is consid- the context of Malaysia today. These policies are ered, with survey data indicating the importance summarized in table 5.1 and elaborated in the of linking popular spending with the taxes need- following subsections. They focus on five main ed to finance them; (4) spatial inequalities; and areas: (1) achieving higher labor incomes, by in- (5) some new indicators are proposed, for mea- creasing workers’ productivity, and addressing suring progress, while additional areas for future human capital gaps earlier in the life cycle; (2) research are suggested. TA B L E 5.1 POLICY MEASURES FOR REDUCING INEQUALIT Y AND P R O M O T I N G E C O N O M I C M O B I L I T Y I N M A L AY S I A Theme Policy measure Enhancing Increase productivity and make better use of underutilized labor opportunities • Large numbers of Malaysians work in jobs that do not make full use of their education and skills. Investment policy needs to attract investments in new and existing enterprises that will create more skilled jobs, especially high-value-added jobs in the services sector. It is crucial that the new progressive wage policy be backed by improvements in productivity. • Current policies to promote access to public sector employment by Bumiputera miss the fact that these jobs are accessible only to a select few, often better-off Bumiputera. Given the limited scope for the public sector to absorb qualified graduates regardless, focusing on broadening the availability of and access to productive private sector employment by all Malaysians, including less well-off Bumiputera and those in East Malaysia. • Increasing women’s labor force participation would promote inclusive growth. Greater availability of flexible work arrangements, affordable childcare, and family leave can make it easier for women to work outside the home. 77 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Theme Policy measure Improve the quality of education and skills development in areas and among households that lag the national average. • Improvements in the educational system are needed to position the next generation for more productive and remunerative high-skilled employment. These medium- to long-term measures need to be complemented with policies to support new skills development among the current workforce, especially people in low-skills, low-wage employment. • Malaysia has greatly expanded access to basic education and healthcare, but the quality of both varies widely. Closing the quality gaps is good for both growth and equity. Explore the extent to which these currently unmeasured quality gaps drive the unexplained earnings gaps between Bumiputera and other workers. • In education, this can be achieved by better tracking of student learning outcomes and teacher performance. • Monitoring systems need to focus on understanding why some children and young people perform worse than others, both across and within ethnicities and states. Increase investment in broad-based human capital development early in the life- cycle • Adequate healthcare, including antenatal care, and early childhood education (ECE) provide the foundation for a productive life and are far more effective and less expensive than remedial measures later in life. Particularly important are the expansion of high-quality ECE and measures to better understand the causes of and solutions to Malaysia’s high rate of stunting. Strengthening social Increase spending on social protection protection • Malaysia’s spending on social protection trails that of comparator countries, including several lower-middle-income countries. In particular, increased spending on cash transfers is needed. Improve targeting across the income distribution to increase spending efficiency • Malaysia’s social assistance spending has good coverage of target groups, but a significant share of benefits leaks to better-off households. Tightening targeting across the income distribution would increase the poverty- and inequality-reducing impact per ringgit spent. • Better targeting also involves understanding the heterogeneous needs within the bottom 40 percent of the population. Reduce fragmentation and improve administrative efficiency • The current system of multiple programs and implementing agencies is administratively inefficient. Consolidation could simplify the system for both beneficiaries and administrators and lead to a larger share of spending going to benefits. Financing inclusive Increase investments in education and health for accelerated and inclusive investments growth • Malaysia spends a smaller share of its budget on education and health than comparator countries. Increased spending in these areas is needed to both support the necessary human capital investments for equitable growth and cope with the health needs of an aging population. Create the fiscal space for greater equity-improving investments • Malaysia’s personal income tax system is progressive but collects relatively little revenue. Reforms can be implemented to enhance collections by broadening the base while maintaining its progressivity. • Untargeted subsidies largely benefit richer individuals. These should be phased out and a portion of the savings channeled to social assistance of targeted beneficiaries. • Consider reintroducing a Goods and Services Tax (GST) or Value Added Tax (VAT), with limited exemptions to increase revenues. Adverse impacts on poorer households could be offset by exemptions or zero-rating of basic necessities and/or a targeted rebate. Communicate the benefits of the package of equity-enhancing policies to the public • Malaysians are concerned about inequality and want the government to act on it. Explicitly tying expenditure policies that generate revenues to the accomplishment of equity-enhancing investments can increase public support for a package of reforms. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 78 Addressing spatial Use place-based and place-sensitive policies to address spatial inequalities inequalities • Location is an important component of inequality in Malaysia. Service quality in East Malaysia and parts of the eastern peninsula lags that in the western peninsula. Investment and spending need to take account of the fact that achieving a level playing field across regions requires spending more in remote areas, even recognizing that the density of services may be lower in remote areas. • Sectoral transformation to high value-added activities may be more limited in lagging areas. The emphasis in these areas should be less on diversification and more on getting to the productivity frontier of existing activities. Policies that facilitate easier internal migration would also be useful. Monitoring Track inequality using a suite of measures inequality • The Gini index provides only a partial understanding of inequality. Adding measures such as the Prosperity Gap and decile shares of income can enhance the understanding of inequality. • Looking at inequality of post-fiscal incomes in addition to market incomes would provide insights into the extent to which fiscal policy reduces income inequality. Systematically track economic mobility as well as inequality • Low economic mobility compounds the negative effects of inequality. Monitoring mobility can be enhanced by (a) adding selected retrospective questions to the HIES/ BA, such as place of birth and educational attainment of the respondent’s parents, and (b) considering adding a longitudinal (panel) component for a subset of the HIES/BA to develop better measures of mobility. Measure the quality of services • Most of the inequality in public services in Malaysia is attributable to differences in quality, not lack of access. Narrowing gaps requires systematic collection of data on quality, which could come from a mix of administrative and survey data, and rigorously evaluating policies designed to improve quality. Strengthen the statistical system • Better monitoring of inequality requires strengthening the statistical system, particularly in terms of data infrastructure. 79 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 5.1 INCREASING OPPORTUNITIES F O R A L L M A L AY S I A N S S T RONG E R R E T UR NS F OR SK IL L S A ND HIG HE R E DUC AT ION CAN DRIVE GROWTH WHILE PROVIDING INDIVIDUALS WITH A KE Y CHANNEL FOR UPWARD MOBILIT Y F or Malaysia to fulfill its laysia’s GDP per capita needs to grow at 5 percent high-income country (HIC) annually, which in turn requires roughly twice the aspirations, increased pro- growth in worker productivity compared with ductivity is needed to rein- what the country has achieved in the previous vigorate growth. While this three decades. Achieving this involves removing report is focused on the mi- barriers to fair competition, making better use croeconomic determinants of of underutilized labor, especially that of women income inequality, the macroeconomic setting of and underemployed young skilled workers, and economic growth and job creation is key for re- adapting and improving the education system so ducing inequality. At the same time, the policies that workers are prepared for the jobs of the fu- needed to do so promote both growth and equi- ture and acquire the skills needed by employers ty when they are wide in scope and relevant for (World Bank 2023a). An example of how these the entire population. The “Aiming High” flagship policies, when broad reaching across the popula- report outlined that, while past growth was pri- tion, promote growth as well as equity is seen in marily due to factor accumulation, productivity an OECD study (OECD, Hanushek and Woess- growth is key to reinvigorating Malaysia’s growth mann 2015) that shows the impact on long-term model and fulfil its HIC aspirations (World Bank growth of (i) increasing education quality (prox- 2021a). Growth projections for the next 25 years ied by rising PISA scores) while keeping current show a mere 2.25 percent annual growth of per enrolment rates constant, and (ii) increasing sec- capita GDP if labor force participation and em- ondary enrolment to reach universal levels but at ployment, and value added per worker (a mea- current quality. For Malaysia, improving quality sure of productivity) remain the same as they of education would rise GDP per capita by 0.53 were just before COVID-19, along with slow points, universal enrolment at current quality aging of the population (figure 5.3). A 2 percent- would only do so by 0.18 points. This points to the age point increase in female participation every importance of rising skills and education quality five years would help, which would push annual for both growth and for inequality, since the qual- growth to 2.5 percent. However, to achieve the ity (scores) are lower for poorer children. average GDP per capita currently for HICs, Ma- FIGURE 5.3 G R O W T H D E C O M P O S I T I O N F O R M A L AY S I A : ( A ) 1991–2021 AND (B) PROJECTIONS FOR 2021–50 6% 5% GDP per capita Growth decomposition (percent) 4% Demographics 3% Employment 2% Participation 1% Productivity 0% -1% -2% -3% 1991-2001 2001-11 2011-19 2019-21 2021-50 High FLFP HIC Target Baseline Source: World Bank Poverty and Inequality Platform (https://pip.worldbank.org/) and World Development Indicators. Note: Growth decomposition based on the World Bank’s Job Structure Tool. See Muller (2008), Gutierrez et al. (2007), and World Bank (2009a) for more details. Scenario projections use UN-projected total and working age populations and hold employment constant. The baseline uses the average worker productivity growth and labor force participation for 2011–19. High FLFP increases FLFP by 2 points every five years. For HICs, high FLFP growth as well as the productivity growth required to achieve the 2022 average HIC GDP per capita by 2050 is used. FLFP = female labor force participation; GDP = gross domestic product; HIC = high- income country. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 80 And inclusive growth will ensure all Malay- which aims to raise the wages of low-income sians reach a high-income living standard. workers as they become more productive, could The prosperity gap (PG) measure showed that a be more effective but will require improvements quarter of Malaysians crossed the high-income in the quality of education and skill development threshold since 2004. It also showed that the to boost workers’ productivity. Meanwhile, exist- remaining three-quarters of the population is ing policies designed to address risk for informal only just over halfway there,61 and, if economic workers are steps in the right direction but have growth is distributionally neutral, in other words, limited coverage so far (Ghorpade et al. 2024). if inequality remains unchanged, then about half In the medium to long term, education and skill would still not reach such an income level by the enhancement to boost workers’ productivity are time the country crosses the HIC threshold in a key complementary policies to increase workers’ few years. More inclusive and pro-poor growth compensation. is needed for every Malaysian to reach a high-in- come living standard. But for this to happen, faster creation of high-skilled jobs is also needed. An enhance- While a decline in the high-education premi- ment in workers’ skills and productivity will only um has contributed to a decline of inequality, result in greater skill underemployment and fur- a continuation of this trend would adversely ther declines in skill premiums unless there are affect Malaysia fulfilling its HIC aspirations. high-skilled jobs for them to engage in. Thus, The declining skill premium has contributed to a demand-side factors, which largely rely on mac- decline of inequality. This is not unusual; for in- roeconomic patterns, are also important for in- stance, the decline in inequality in various Latin equality and mobility. Structural transformation American countries contributed to a falling skill from lower- to higher-value-added activities is premium as the relative supply of skilled workers an essential driver of growth. In fact, Malaysia’s rose (e.g., Ferreira et al. 2008; Barros et al. 2010; transition from a low- to middle-income econo- Lustig, López-Calva, and Ortiz-Juárez 2013; Gas- my was closely linked to a rapid transition from parini et al. 2011). But ultimately, inequality and an agricultural to an industrial economy (World growth would both be supported if there are Bank 2023a). While the economic structure higher returns to skills that are strongly linked to of Malaysia is starting to shift toward high-val- productivity. ue-added services, the employment structure remains more reliant on low-skilled services. The Higher labor income returns for poorer high rate of skill-related underemployment in the households will help decrease wage inequal- country is in part a reflection of the lack of these ity but need to be driven by higher productiv- jobs for workers when they have higher educa- ity. Minimum wages are often discussed as a po- tional levels. For instance, in Mexico, the aggre- tential instrument for reducing income inequality, gate resource misallocation toward less produc- but these policies have limited ability to reach the tive firms, which employ less highly educated most disadvantaged workers and do not facilitate workers, contributed to a stagnation of earnings either physical or labor mobility. International for workers and to a decline in the skill premium experience illustrates this point. For instance, (Levy and López-Calva 2019). Thus, inclusive Ferreira, Firpo, and Messina (2017) find that ris- growth requires creating jobs in high-value-add- ing minimum wages contributed to a decline of ed sectors where increasing numbers of educated inequality in Brazil in 2003–12 but not in 1995– Malaysian workers can go. This does not need to 2003, when limited compliance with the policy drive up inequality. On the one hand, the most led to higher wage gaps between workers who lucrative jobs are likely to constitute a small num- kept formal sector jobs and those who did not. ber—not enough for a sizeable rise in inequality. Malaysia introduced minimum wage in 2014, but Further, for inequality to decrease, workers from it is unlikely to benefit the majority of those at the poorer households also need the same chance bottom of the income distribution—even though to gain these higher-paying and higher-skilled its purpose is to raise the minimum income re- jobs. This means they need to have had the same ceived by workers—because they are much more chance to accumulate quality human capital as likely to be in informal employment and in low- everyone else, a topic addressed next, as well as skilled and low-productivity occupations. The to take risks as and when productive opportuni- recently announced “progressive wage” policy, ties arise. 61. Eliminating the PG in Malaysia would require a 1.8 times growth in the average income for those below the high-income line. 81 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA I N C R E A S I N G O P P O R T U N I T I E S F O R A L L E A R LY O N I N L I F E AND THROUGH A FOCUS ON SYSTEMIC IMPROVEMENTS IN H E A LT H A N D E D U C AT I O N There are already several policies and pro- them, regardless of their ethnic background, the grams to tackle inequality in Malaysia; they most cost-effective entry points for interventions could strengthen the emphasis on address- can be identified by focusing on the compound- ing today’s needs. Inequality reduction features ing reasons for their falling behind. Meanwhile, prominently in Malaysia’s major planning and employment policies focusing on preferential ac- strategy documents (e.g., the Twelfth Malaysia cess to government jobs miss the fact that these Plan and its mid-term review and the recent- opportunities are accessible only to a select few, ly released MADANI framework). Historically, often better-off Bumiputera. The public sector’s significant emphasis has been placed on closing limited capacity to absorb increasingly qualified ethnic gaps through affirmative action policies. labor market entrants regardless, means that in Notable policies include (1) increasing educa- today’s labor market, the quality and relevance tional attainment; (2) facilitating access to em- of degrees granted to, and the soft skills acquired ployment opportunities (particularly through the by, graduates become more important in terms of public sector62); and (3) some support for micro, gaining productive and high-skilled private sec- small, and medium enterprises. Among the pro- tor jobs. Further research is needed to understand grams specifically for building human capital for why there is an income disparity between some the Bumiputera, the main is the 90 percent quota Bumiputera workers and others who have similar in colleges, which exclusively covers Bumiputera education and other attributes. Any complemen- MARA institutions and scholarships, while the tary policies that do focus on the reasons for any Yayasan Peneraju Pendidikan Bumiputera (YPPB) further ethnic gaps, need to ensure they benefit scholarships for technical/professional training the disadvantaged Bumiputera rather than those for disadvantaged students is a smaller, niche pro- who are already succeeding. gram. Because inequality of outcomes has roots Emphasis on access gap closure may have early in life, addressing inequality for the fu- contributed to inequality reduction in the ture generation must emphasize increasing past, but it may not be sufficient to address these opportunities for all Malaysians. Given the gaps that remain today. Past policies may large gaps in health and education access have have helped narrow access gaps, especially in mostly been closed, addressing the quality gap terms of increased tertiary educational attain- has risen as an increasingly important policy fo- ment, but that may not be sufficient or appropri- cus for improving human capital development, ate going forward for Malaysia as a high-income for the lagging Bumiputera as well as for all other and developed economy (World Bank 2023a). Malaysian children. This means focusing on sys- Given enrollment rates and educational attain- temic policies that are broad reaching. For edu- ment have caught up for many disadvantaged cation, the most recent Malaysia Economic Mon- children, this is no longer the main gap. Nonethe- itor (World Bank 2024a) highlights the following less, there are still a few youth who are lagging; for policy areas to improve the education system so 62. Also, procurement in government contracts. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 82 that foundational skills are developed eventually: long impacts and ultimately lead to unequal re- (1) reaching the last mile to ensure free and com- turns in the labor markets. Early childhood edu- pulsory preschool education to give all children cation, for instance, fosters cognitive skills along a good head start; (2) rigorously and more fre- with soft skills such as attentiveness, motivation, quently measuring student learning and teach- self-control, and sociability, which are as import- er performance and strengthening the teacher ant for educational outcomes further on, as well appraisal system; and (3) incentivizing teachers’ as for performance on job. Further, the return on performance through training programs that em- the investment in developing early skills is high phasize content knowledge, practice with col- and this is a more cost-effective strategy than ad- leagues, follow-up support, and career incentives, dressing bigger gaps later in life (Heckman et al. but also by integrating teachers’ experiences into 2010; figure 5.4). For instance, the US Perry Pre- the design and implementation of policies for school program shows a 7–10 percent return on teachers. Regarding health, there is a need to bet- investment per year based on increased school ter understand what causes high stunting rates in and career achievement, as well as reduced costs the country and what can be done about them. for remedial education, health, and expenditures related to the criminal justice system (Heckman Strategies focusing on early-life investments et al. 2010). Similarly, the Abecedarian/CARE’s that address foundational gaps are more ef- birth-to-five early childhood programs for disad- ficient and more cost-effective. Focusing on vantaged children yielded a 13 percent return on access to higher education or facilitating public investment per child per year (Garcia et al. 2017). sector jobs through quotas or preferential access Interventions through social safety net programs not only tend to benefit the better-off, who have for families with young children also yield higher made it through the educational system, they are gains for younger than older children in terms of also more expensive strategies compared with their impact outcomes measured in later adoles- focusing on early childhood education and skills. cence and into adulthood (Chetty, Hendren, and Addressing inequality of opportunity requires Katz 2016). addressing the foundational gaps that leave life- F I G U R E 5.4 RETURNS ON A UNIT OF DOLL AR INVESTED Rate of return to investment in human capital Programs targeted toward the earliest years Preschool programs Schooling Job training 3 5 al ol l oo – – at ho ch 0 4 en Sc -s Pr st Po Heckman. James J. (2008). “Schools. Skills and Synapses.” Economic Inquiry. 46(3): 289-324 Source: Heckman (2008). 83 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 5.2 STRENGTHENING SOCIAL PROTECTION TO ADDRESS THE NE E D S AT T HE BO T T OM OF T HE INCOME DISTRIBUTION T he Malaysian system of taxes Malaysia comes from health and education ben- and transfers is progressive efits, which, although important for mobility and and reduces inequality, but long-term prosperity, do not help poor house- it trails many upper-mid- holds meet their daily living needs. Cash transfers dle-income country (UMIC) are better suited for this task, but in Malaysia the peers and even some low- high fragmentation and administrative inefficien- and middle-income coun- cies of the social protection system, coupled with tries (LMICs) in the inequality reduction that substantial inclusion error and social assistance it achieves. Taxes and transfers can help mitigate that does not target the poor  (figure 5.5, Panel shocks and support households through the in- A) and a relatively low level of public spending come-generating process. Taxes, cash transfers, on social assistance (World Bank 2023a) means and subsidies in Malaysia reduce inequality by that only about two-thirds of these transfers go to 2.4 points from the pre-fiscal level, which is close the bottom 40 percent (Panel B), and the amount to the UMIC average. When noncash health and paid is too low (around 13 percent of income for education benefits are included, Malaysia’s in- the poorest 10 percent of Malaysians and 7 per- equality reduction is 4.1 points higher, for a total cent for the second-poorest 10 percent) (Panel of 6.5 points. While the fiscal system achieves C). There were some modest improvements be- some inequality reduction and is progressive, it tween 2019 and 2022, with fewer richer house- trails many UMIC peers. It ranks 18th out of 25 holds getting benefits (Panel A) and all households UMICs—10 LMICs perform better than Malaysia. getting slightly higher value (Panel C) as program benefits increased for the headline cash transfer The Malaysian categorization of households program, Sumbangan Rahmah Tunai  (SRT) and into the bottom 40/middle 40/top 20 percent additional cash and noncash support was given results in groups that are heterogeneous. The through nonstandard programs.63 traditional Malaysian categories used to analyze outcomes across the income distribution, that is, Meanwhile, subsidies are also used to sup- the bottom 40/middle 40/top 20, result in groups port poorer households, but they cost much that are heterogenous and not well suited for more than social assistance and have a small- policy design. For instance, bottom 40 percent er impact on poverty. In 2022, public spending has impoverished individuals, who need basic in- on subsidies alone was 2.9 percent of the gross come support; vulnerable individuals, who need domestic product (GDP); meanwhile, public protection to better cope with risks, for example, spending on social assistance was just over 1 per- though social insurance policies for the informal- cent (World Bank 2023a). However, the share of ly employed; and individuals who are not as vul- fuel subsidy benefits, for example, those for food nerable but are yet not economically secure and and agriculture, is higher for richer than poorer need policies that may let them save and invest in households (Figure 5.6). Beyond discouraging en- riskier but also potentially more rewarding eco- ergy efficiency and contributing to carbon emis- nomic activities. sions, fuel subsidies are only a quarter as effective at reducing poverty as targeted cash transfers per Social assistance for poorer households is ringgit spent, do little to reduce inequality, and fragmented and spread over far too many represent a significant opportunity cost in terms households, including many that do not need of budget allocation (World Bank 2023a). it, diluting the benefits for those that do. Most inequality reduction through the fiscal system in People-based policies should focus on those 63. A new World Bank study on social assistance program outcomes and exclusion in 2019 and 2022 is expected in early 2025. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 84 F I G U R E 5.5 C O V E R A G E , S H A R E A N D VA L U E O F S O C I A L A S S I S TA N C E B E N E F I T S B Y D E C I L E , 2 0 1 9 A N D 2 0 2 2 Panel A: Coverage by decile Panel B: Share of total Panel C: Value of benefits by decile (receives any program) benefits by decile (all programs). relative to income 100 30 15 80 20 10 60 Percent 40 10 5 20 0 0 0 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Pre-transfer income decile Pre-transfer income decile Pre-transfer income decile 2019 2022 Source: World Bank staff calculations based on data from DOSM HIES/BA. Notes: Household per capita income deciles are based on “pre-transfer income”, which is gross household income less transfers. This is similar to but slightly different than “market income” which makes further income tax adjustments and is used in World Bank (2023a). Results to not differ materially. who need support most, wherever they are ceives, as indicated earlier. A more pro-poor fiscal and regardless of their ethnic group or oc- system without additional net spending could be cupation. In most countries, the gap between achieved by a series of reforms that redirect subsi- current coverage and the goal of Universal Social dies into more progressive transfers. For example, Protection (USP) will not be closed overnight. In a simulation exercise64 in which the fuel subsidy this context, one path towards USP is to prioritize (2.9 percent of GDP in 2022) was eliminated65 those households in greatest need first (Grosh, and less than 20 percent of the proceeds were Leite, Wai-Poi and Tesliuc 2022). In fact, most used to expand current cash transfers achieved poorer households in Malaysia already receive the same poverty and inequality reduction at a some form of government assistance (85–97 per- much lower cost than what the country spent cent of the Bottom 40 receive support). However, on fuel subsidies (Figure 5.7). Ensuring more so- 56–80 percent of the Middle 40 and 33 percent cial assistance spending reaches the B40 would of the Top 20 do so as well. As a result, only 65 have even larger poverty and inequality impacts percent of transfers go to the Bottom 40. Given (World Bank 2023a). the current budget constraints, this significantly limits the benefit amount that each household re- Increasing financial literacy and inclusion F I G U R E 5.6 SHARE OF SUBSIDIES BY DECILE, 2019 Share of subsidies 20 15 Percent 10 Food & Agriculture 5 All Subsidies Fuel 0 1 2 3 4 5 6 7 8 9 10 Per capita income decile Source: World Bank (2023a) and World Bank staff calculations based on DOSM HIES/BA (2019). Note: Estimates adjusted to the baseline of the fuel subsidy spending in 2022. Scenarios: (1) removing the fuel subsidy and (2) increasing the social assistance budget to 1.5 percent of GDP, with higher benefit levels and existing targeting. SA = social assistance. 64. See World Bank (2023a) for details of the scenarios and methodology. 65. In October 2024, the government announced a commitment to roll back petrol (RON95) subsidies for the wealthiest 15 percent of the population in mid-2025. 85 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE 5. 7 P O V E R T Y A N D I N E Q U A L I T Y R E D U C T I O N (G I N I I N D E X ) A N D F I S C A L S AV I N G S ( P E R C E N T O F G D P ) U N D E R F U E L S U B S I D Y A N D S O C I A L A S S I S TA N C E REFORM SCENARIOS 3 3 2 2 1 0 -1 1 -2 -3 0 -4 -5 -1 -6 -7 -8 -2 2022 Remove fuel. 2022 Remove fuel. fuel subsidy Increase SA fuel subsidy Increase SA Poverty impact Net fiscal savings Cost-effectiveness Inequality impact Source: World Bank (2023a) and World Bank staff calculations based on DOSM HIES/BA (2019). Note: Estimates adjusted to the baseline of the fuel subsidy spending in 2022. Scenarios: (1) removing the fuel subsidy and (2) increasing the social assistance budget to 1.5 percent of GDP, with higher benefit levels and existing targeting. SA = social assistance. can help households build financial resilience crofinance and microinsurance products such as and the ability to cope with shocks. Bank Neg- iTekad and Perlindungan Tenang have been made ara Malaysia (BNM) has launched initiatives in its more available. Complementing these strate- Financial Sector Blueprint 2022–26 and Finan- gies, the National Strategy for Financial Litera- cial Inclusion Framework 2023–26 to increase cy 2019–2366 also sets out initiatives to increase financial inclusion, especially for the financially financial literacy and access to financial knowl- unserved and underserved. Some of the key mea- edge among the financially unserved and under- sures include encouraging financial service pro- served. It includes initiatives to increase financial viders to create tailored products for low-income education, management, consumer protection, consumers, considering their unique risks and and collaboration among sectors. Additionally, challenges and promoting digital finance tools support systems for diverse groups, including gig to extend services to underserved communi- workers and rural residents, are being developed ties while ensuring consumer protection against to promote resilience and financial planning. predatory practices. Through these measures, mi- 66. A new version is currently being developed. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 86 5.3 ADDRESSING SPAT I A L INEQ UA L I T IE S M Many of the remaining (World Bank 2022a); electricity outages in Sabah ethnic gaps are tight- are thrice as long as in Peninsular Malaysia; and ly linked with location; many firms in this state report difficulties with this highlights the role other infrastructure services such as water, trans- for place-based and portation, and internet coverage (World Bank place-sensitive policies. 2022a). Policies to bring services in lagging areas While ethnic gaps have been up to par with those in leading areas need to aim closed considerably, those that remain are tightly for equity in outcomes as opposed to equality of linked with location. While spatially blind pol- public expenditure. For example, experience in icies, for example, investing in “portable human the United States has shown that achieving parity capital,” are important for inclusive growth and in education requires higher spending per stu- can facilitate labor migration, there is also a role dent in lagging areas, to not only make up for the for policies that are based on and sensitive to existing gap, but also because services are gener- place. Placed-based policies (see Rodríguez-Pose ally more expensive to deliver when the lagging et al. 2024) in the lagging regions can be used to areas are more sparsely populated, as they are in support the Bumiputera in the East, as well as the Malaysia (World Bank 2009b). Likewise, because impoverished in other lagging regions. Malaysia’s lagging areas have more difficult ter- rain, along with a smaller user base than urban Policies to invigorate the local economy, centers, it will cost more to extend broadband to guided by their comparative advantages, the lagging areas help close the digital divide than should form the core of efforts to help lag- to deliver it to urban centers.67 ging regions close the gap. Public spending and investment in human capital and physical capital, The scope for sectoral transformation to such as infrastructure, need to be strategic, with a high-value-added activities may be more lim- view to sustainably raising living standards by de- ited in lagging areas. Malaysia’s lagging regions veloping new businesses and creating good jobs have not experienced the same level of econom- through private sector development. An example ic transformation as elsewhere in the country. is the European Union (EU), which allocates al- Sabah, for instance, did not experience as much most a third of its budget to supporting its Cohe- the manufacturing boom, which facilitated the sion Policy with an objective of reducing dispari- movement of labor out of agriculture in the rest of ties both between and within EU Member States. Malaysia (World Bank 2022a). In such cases, the European Commission partners work together emphasis should be less on diversification, which with national and subnational governments would mean skipping some necessary stages of through a set of specific funds to identify and im- the structural transformation process, and more plement investments that spur economic growth on getting to the productivity frontier of the ex- in lagging regions through innovation, improved isting activities through investment for adequate competitiveness, attention to sustainability (es- physical infrastructure and qualified human cap- pecially environmental), and improved integra- ital. Guiding these investments requires a more tion in the national and international economies detailed analysis of the constraints to improve (European Commission 2017). agricultural productivity in the lagging regions of Malaysia. Finally, there is also a role for policies Improving public services in lagging regions that promote connectivity and facilitate easier is a necessity. Service quality in East Malay- internal migration within states such as Sabah, sia and parts of the eastern peninsula lags that Sarawak, and Kelantan, and from these places to in the western peninsula. For instance, educa- the rest of Malaysia. tion quality is worse in Malaysia’s lagging areas 67. Some programs already being rolled out, for example, the Pelan Jalinan Digital Negara (JENDELA), support sustainable, inclusive, and high-speed broadband connectivity, especially in rural and remote areas. 87 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 5.4 FINANCING INCLUSIVE INVESTMENTS F inancing the expenditures the rise of noncommunicable diseases. Educa- needed for Malayasia to tion and health budgets are lower as a percent- fulfil its HIC ambitions will age of GDP than in other countries (figure 5.11). cost it more than it current- The current health budget (2 percent of GDP) is ly spends. In particular, the far lower than transitional and aspirational peers’ amount of spending on so- (the UMIC average is 4 percent, and the average cial protection and health for the countries in East Asia and Pacific [EAP] and education in Malaysia is not sufficient is 4.9 percent), while education spending, as a and lower than in benchmark countries. To percentage of GDP, has been declining, from 5.7 fulfil its HIC ambitions, Malaysia will need to percent in 2012 to 3.5 percent in 2022, the low- support greater investments in human capital est in two decades (Figure 5.8). This is despite and productivity-boosting physical and digital education representing 20 percent of all govern- infrastructure. Among others, both health and ment spending—a proportion that has remained education will need further investments: edu- constant over time; the declining real spending is a cation quality is lagging, and health spending consequence of falling total revenues and spending. needs will increase as Malaysians age, and with FIGURE 5.8 E D U C AT I O N A N D H E A LT H S P E N D I N G , 2 0 0 4 – 2 2 Percent of government expenditure 7 25 Education 6 (% gov expenditures. 20 5 RHS) Percent of GDP 4 15 Health 3 10 (% GDP) 2 5 Education 1 (% GDP) 0 0 2004 2007 2010 2013 2016 2019 2022 Source: World Development Indicators. Note: GDP = gross domestic product; RHS = right hand side. Fundamentally, Malaysia’s low public reve- concentrated and still constitute a high share of nues limit the scope for greater public spend- the national income, are not taxed, and direct tax- ing. Malaysia’s revenues not only lag the revenues ation through PIT is low, at below 3 percent of of UMIC and HIC peers, but also the LMIC av- GDP. PIT is also designed in a manner that plac- erage (Figure 5.9). The country thus needs to es an inherent limit on its revenue capacity and raise more revenues through both direct and compromises the progressivity of the tax burden. indirect taxes. The biggest consequence of low The following are three core features of the de- revenues is that they limit what a country can sign that can be improved: (1) the relatively high spend. Moreover, international data suggest that chargeable income thresholds and low tax rates running a government has an associated relative- in the upper-income brackets, (2) the availability ly fixed cost, and total spending in addition to of multiple reliefs with no overall cap, and (3) the that on health, education, and social protection relatively narrow scope of the tax (World Bank is relatively fixed as a percentage of GDP across 2023a). Addressing these features would broaden all income levels (Figure 5.10). This means that as the tax base while mostly affecting taxpayers in new revenues are raised, fiscal space is created for the higher income deciles, and, at the same time, greater social spending. drive up revenues. Estimates based on Malaysia’s PIT Policy Microsimulation Model suggest that On the revenue side, there is a role for fair lowering the taxable income thresholds and ap- taxation. The design of personal income tax plying higher rates in the upper-income brack- (PIT) could be improved to make it more pro- ets could increase PIT revenue up to RM 2.5–2.8 gressive and enable it to address inequality at billion in the year of assessment, 2024, while im- the top of the income distribution. Malaysia’s posing an overall cap on the reliefs claimed could heavy reliance on direct taxation is quite pro- add up at least another RM 1.1 billion. gressive. But capital incomes, which are highly A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 88 F I G U R E 5.9 45 GOVERNMENT 40 REVENUE BY SOURCE, 35 PERCENT OF GDP 30 25 Source: World Bank 2022c, 2023a. 20 Note: GDP = gross domestic product; HIC 15 = high-income country; LIC = low-income 10 country; LMIC = lower-middle-income 5 country; OECD = Or- ganisation for Econom- 0 ic Co-operation and LIC LMIC UMIC Malaysia HIC OECD Development; UMIC = upper-middle-income country. Non-tax Indirect Direct F I G U R E 5.1 0 45 GOVERNMENT 40 E XPENDITURE BY T YPE, PERCENT 35 OF GDP 30 25 20 15 Source: World Bank 2022c, 10 2023a. Note: GDP = gross 5 domestic product; HIC = high-income country; LIC 0 = low-income country; LIC LMIC Malaysia HIC OECD LMIC = lower-middle-in- come country; UMIC = Other Pensions Social Protection upper-middle-income (ex. Pens.) country. Health Education Phased fiscal reforms could generate new sidies do not create a sustainable source of new revenues, allow increased development fiscal space; the size of the subsidies depends on spending, and reduce poverty and inequal- international fuel prices, but so does much of Ma- ity. The first phase would be to remove fuel laysia’s revenues. Thus, new and sustainable reve- subsidies and compensate poorer households nues are required. with increased social assistance. The Octo- ber 2023 Malaysia Economic Monitor outlined Phase 2 would significantly increase long- a reform scenario where costly and regressive term revenues through both indirect and fuel subsidies could be removed (saving over 2.5 direct taxation. Malaysia’s indirect tax collec- percent of GDP in spending), while redirecting a tion, at 3 percent of GDP, is particularly low, small proportion into increased social assistance below even the LIC average. Thus, in the short (2023c). In the current report, this is taken as a term at least, increased revenue collection can first phase of fiscal reform—bringing social assis- be achieved through indirect taxation. Consid- tance spending up from 1.0 percent of GDP to 1.5 eration should be given to the reintroduction of percent, closer to the 1.8 percent UMIC average. the goods and services tax (GST) (or value added As shown in the previous section, even if social tax), since it is widely regarded as one of the most assistance is increased but with the existing tar- efficient taxes in a country’s toolkit and can be geting, poverty would decline slightly. However, implemented widely and swiftly. A simulation ex- the fiscal savings due to the removal of fuel sub- ercise shows that replacing the current sales and 89 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA services tax (SST) with a GST with a 10 percent Part of the increased revenues could finance base rate (roughly the EAP region benchmark) an increase in health expenditures. Spending with few exemptions or preferential rates would on health could be increased by 1 percent of GDP. generate additional revenues constituting about Even if the benefits of this increase are only pro- 1 percent of GDP, with relatively little impact on portional to the existing health benefits received inequality (figure 5.12). Moreover, the estimated across the income distribution, this would reduce impact of consumption taxes may be even more inequality further by 1.3 points, since the same progressive because poor households have a level of benefits means much more to a poorer higher degree of informal consumption, which household as a percentage of their much lower attracts less tax (Bachas, Gadenne, and Jensen income. Some education-related reforms (see 2023).68 In addition, at least an extra 1 percent of World Bank 2024a could be implemented to im- GDP could be raised from increased PIT collec- prove outcomes within the existing budget. tion due to the specific reforms outlined above— this would not only create additional fiscal space, Increased revenues can also be used to but would reduce inequality directly by being support a “just transition.” Even though the paid mostly by richer households (while keeping future benefits of a green transition—toward a the overall income tax burden reasonable; World low-carbon, climate-resilient economy—will Bank 2023a). be shared by all, specific individuals may face higher short-term costs (Hill, Nguyen, and Khanh The burden on poorer households scan be Doan 2024). In particular, costs such as higher offset through targeting transfers better and energy prices or job losses in carbon-intensive introducing a targeted GST rebate. Still, since sectors can be especially hard to manage for the any increase in indirect taxes will place some bur- impoverished (Hill, Nguyen, and Khanh Doan den on the poorer households, there should be 2024). Additional revenues will be needed to measures to mitigate this impact. Better targeting finance measures to compensate those affected of social assistance toward the poorest house- by fuel subsidy removal, as shown previously, holds in the bottom 40 percent, along with a tax but also to support poorer Malaysian households rebate, would offset the impact of a GST increase to make the productive investments needed to on poverty, have no impact on inequality, and still transition.69 leave additional fiscal savings equivalent to 2 per- cent of GDP (Figure 5.11). F I G U R E 5 .1 1 I N E Q U A L I T Y R E D U C T I O N (G I N I I N D E X ) A N D F I S C A L S AV I N G S ( P E R C E N T O F G D P ) U N D E R D I F F E R E N T S C E N A R I O S O F F U E L S U B S I D I E S , VA L U E A D D E D , A N D S O C I A L A S S I S TA N C E R E F O R M S , R E L AT I V E T O B A S E L I N E G S T S I M U L AT I O N S 4 2 0 –2 –4 –6 –8 –10 2022 fuel Remove fuel. +GST +GST, rebate and +PIT and health subsidy increase SA (no exemptions) improved targeting SA to B40 Inequality impact Net fiscal savings Cost-effectiveness Source: World Bank (2023a) and World Bank staff calculations based on the DOSM HIES/BA. Note: Scenarios: (0) baseline of the 2022 fuel subsidy budget; (1) removing the fuel subsidy and increasing the social assistance (SA) budget to 1.5 percent of GDP with higher benefits and existing targeting; (2) replacing the existing SST with a GST with a 10 percent base rate and no exemptions; (3) adding a tax rebate for poorer households (in the B40) and increasing the SA budget to 1.5 percent of GDP, with higher benefits, and improving targeting by restricting coverage to B40 households using PMT targeting only; and (4) adding 1 percent of GDP from a PIT reform to the budget and using it to raise the levels of in-kind health benefits to current beneficiaries. GDP = gross domestic product; GST = goods and services tax; PIT = personal income tax; SA = social assistance. 68. Consumption taxes are not applied to final sales at informal locations, but there can still be an embedded tax on inputs pur- chased formally, such as inventory and utilities. 69. For instance, reskilling and upskilling, facilitating mobility, and reducing market frictions and credit market failures. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 90 These reforms would lead to a significant of a tax rebate for poorer households; increase of improvement of the inequality reduction the social assistance spending to up to 1.5 percent through the Malaysian system of taxes and of GDP while targeting it better, to the poorest; spending. Malaysia’s current system of taxes, and, finally, reformation of the PIT and the use of transfers, and subsidies enables inequality reduc- the additional revenue to finance greater health tion by about the UMIC average in terms of cash investments—would greatly improve Malaysia’s benefits. When noncash health and education fiscal system’s performance in reducing inequal- benefits are included, the benefits are relatively ity. The fiscal system would help Malaysia prog- small compared with that of other UMICs (figure ress from raking in the bottom half to being in the 5.13). The series of reforms outlined—namely, the top half of countries in terms of the fiscal system’s elimination of the fuel subsidy; the reintroduc- inequality-reducing impact (Figure 5.12). tion of a GST without exemptions; the addition F I G U R E 5.1 2 M A L AY S I A’ S I M PA C T O N I N E Q U A L I T Y (G I N I I N D E X ) T H R O U G H TA X E S , TRANSFERS, SUBSIDIES, AND IN-KIND SERVICES: BASELINE AND REFORM SCENARIOS High income Upper-middle income Lower-middle income Low income Dominican Republic Malaysia (reform) Venezuela. RB United States Côte d’Ivoire Burkina Faso Gambia. The South Africa Russian Fed Guatemala El Salvador Costa Rica Nicaragua Argentina Bostwana Indonesia Honduras Colombia Tajikistan Mauritius Mongolia Comoros Paraguay Sri Lanka Tanzania Romania Thailand Malaysia Moldova Equador Eswatini Uruguay Namibia Ethiopia Panama Georgia Lesotho Ukraine Uganda Albania Zambia Belarus Croatia Turkiye Jordan Mexico Guinea Tunisia Bolivia Ghana Kenya China Egypt Spain Brazil Niger Togo India Peru Mali Iran 0 Change In Gini Index (points) -5 -10 Malaysia -15 (baseline) Malaysia -20 (after reform) In-kind spending on health and education -25 Cash taxes and transfers Net fiscal impact Source: World Bank (2023a) and World Bank staff calculations based on the DOSM HIES/BA. Note: H+E = health and education. C OMMUNIC AT ING T HE PACK AG E OF EQUI T Y- E NH A N CING POLICIES TO THE PUBLIC Inequality reduction is not just a matter of tages of the package clearly to the public (and of- identifying policies that work technically; ten a broad range of policymakers and politicians, the policies required to reduce inequality as well as special interest groups).” also need to work politically. The technical solutions to policy objectives are not straightfor- Particular concerns arise regarding targeted ward. Should a fixed social assistance budget be cash transfers, the most cost-effective means used to cover more people with lower benefits or of reducing inequality. Policy makers and the less people with higher benefits? How can educa- public alike often raise concerns about provid- tional quality be improved? What is the right bal- ing cash transfers to poorer households, despite ance between primary and tertiary health care? strong global evidence that these are the most However, the political solutions can be as import- cost-effective means to reduce inequality in a ant and difficult to design. As Sosa and Wai-Poi technical sense (see World Bank 2022c; Sosa and (forthcoming) note: “Key challenges arise in de- Wai-Poi, forthcoming). The two main concerns signing a reform package that has more winners are that the transfers will create disincentives to than losers (or if there are more losers, the degree work or be spent on “bad” consumption such of loss is relatively small and unobjectionable), as tobacco and alcohol, even though there is no and in communicating the rationale and advan- evidence that these are warranted concerns in 91 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA developing countries.70 Moreover, these percep- whether people think that poverty is due to lazi- tions matter. There is a strong correlation be- ness or an unfair society (figure 5.14). tween the public budget for cash transfers and F I G U R E 5 .1 3 SIZE OF THE CASH TRANSFER BUDGET AND PUBLIC PERCEPTION OF THE REASONS FOR POVERT Y 2.5 Ukraine 2 Croatia Cash Transfers (spending % GDP) Estonia Belarus Serbia 1.5 Romania South Africa Pakistan Moldova Armenia Georgia Montenegro 1 Bulgaria Latvia Chile .5 Macedonia Albania Azerbaijan Colombia Bosnia Turkey Bangladesh Uruguay Mexico Peru El Salvador 0 Nigeria India Philippines 0 .2 .4 .6 .8 Poor because lazy (vs. because unfair society) Source: Banerjee et al. (2017) using the World Values Survey. Note: The figure has been constructed using data on beliefs from the World Values Survey (WVS) and data on national spending on social assistance from the ASPIRE data set for the latest available year. The horizontal axis plots the national average answer to the WVS question “Why, in your opinion, are there people in this country who live in need? Poor because of laziness and lack of will power (=1), OR, Poor because of an unfair society (=0)” from the 1995 WVS wave. GDP = gross domestic product. Thus, it is vital to understand public percep- support income support policies, policies tions and account for them when designing reducing the cost of living, and opportuni- and communicating reforms. People are often ty-boosting policies as a means to achieving mistaken about who wins and who loses due to a inequality. The following paragraphs draw on a given reform. For example, in the United States, perceptions survey conducted for this report by survey respondents believe that 20 percent of the Department of Statistics Malaysia (DOSM) households pay the top PIT rate compared with and the World Bank to understand what people the actual share of 1 percent, and that 25 percent think about inequality and mobility as well as of households pay no PIT, when in reality, it is government actions to address both. The current 44 percent (Stantcheva 2021). Moreover, per- survey is the first one to be conducted in Malay- ceptions data from surveys can be invaluable in sia and thus provides critical insights into how helping to design a communication strategy. For inequality is understood by its people. Malaysians example, a study of 12 middle-income countries believe that the labor market policy (higher mini- finds that while baseline support for a fuel subsidy mum wage), policies to support the impoverished reform is low, the support doubles or even triples (better social protection and subsidies), policies when the reform is packaged with compensatory promoting equal opportunities (free quality ed- policies (Hoy et al. 2023). World Bank (2023d) il- ucation and health care), and policies improving lustrates the international experience of fuel sub- policy implementation (corruption reduction) sidy reform as an example of the political econo- are important in addressing inequality (Figure my challenges of fiscal reform.71 5.14). These top five priorities are consistent for people in all income groups, although the de- A majority of Malaysians think it is relative- mand for social protection and higher minimum ly urgent to address inequality. They tend to wage is lower in higher quintiles (Figure 5.15).72 70. See World Bank (2023) for a summary of the evidence. 71. See Sosa and Wai-Poi (forthcoming) for a more detailed discussion on the political economy of fiscal reform. 72. The results are quite similar to those of a similar survey in Indonesia; see World Bank (2015). A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 92 Significant support is retained by tax in- could take leverage the public support for greater crease, which is needed to facilitate the more tax collection and explicitly tie this to the financ- popular policies. Two of the least popular poli- ing of popular policies such as social assistance, cies for inequality reduction are increasing taxes health, and education spending. Subsidy reform for the rich and increasing the SST or GST (Fig- could also be articulated as creating a space for ure 5.14). However, that does not mean they are more progressive spending on health and edu- without significant support. When asked specif- cation. Further, tax increase might receive more ically about income taxes, half of the Malaysians support if concerns about corruption were ad- thought collection should be increased (Figure dressed (Figure 5.15), and effectiveness of the 5.16), and even the rich hold a perception that spending on social assistance, health, and educa- income tax rates should be more progressive tion were improved and demonstrated. (figure 5.18). Thus, a communication campaign F I G U R E 5.1 4 MOST POPUL AR POLICIES FOR REDUCING INEQUALIT Y Raising the minimum wage 44.6% Providing social protection to poor citizens 38.4% Provide more subsidies 24.0% Reduce corruption 19.4% Provide free quality education for all 18.9% Providing quality healthcare that is free for all 12.3% Providing employment or social insurance for people 10.1% who have lost their jobs Build better infrastructure 8.8% Capital assistance for small businesses 8.3% Increasing taxes for the rich 5.4% Fair ownership of assets for the public 5.1% Increase goods and services tax 2.5% Loans for the poor 2.3% Percent of respondents Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the following options, state two of the most important things that the government needs to do to reduce inequality in Malaysia? 93 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E 5 .1 5 MOST POPUL AR POLICIES FOR REDUCING INEQUALIT Y A. By perceived income class 45% 44% 43% 32% 25% 21% 19% 17% 14% 15% Raising the Providing social protect Provide more subsidies Provide free quality Providing quality minimum wage -ion to poor citizens (e.g. for fuel. food. education for all healthcare that is (such as cash assistance. agriculture. etc.) free for all scholarships. public health insurance. employee insurance. etc.) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 B. By actual income class 50% 49% 39% 29% 24% 25% 24% 17% 18% 15% Raising the minimum Providing social protect Provide more subsidies Provide free quality Reduce corruption wage -ion to poor citizens (e.g. for fuel. food. education for all (such as cash assistance. agriculture. etc.) scholarships. public health insurance. employee insurance. etc.) Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the following options, state two of the most important things that the government needs to do to reduce inequality in Malaysia? A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 94 The political economy of reform also in- role of taxes, including taxes that are regressive cludes communicating to Malaysians how or neutral, is primarily to raise the revenues need- the government strategy is tackling inequal- ed to finance other policies. Often, the revenues ity. Malaysians clearly care about inequality and raised are more than enough to compensate the want the government to act on it. Yet, many of poorest affected households and leave enough the policies they wish to see adopted require to save or invest in equity-increasing policies. A fiscal spending, whereas some others are not as policy package could have more public support if effective as people think they are. For example, people are aware of this complementarity. For ex- in many countries, exemptions and reductions ample, a communication campaign could lever- on the consumption tax for basic products often age the public support for increased tax collec- benefit the richest the most and so do consump- tion and explicitly tie this to the financing of the tion subsidies (for electricity, gasoline, water, etc.) more popular policies such as investment in im- (Inchauste, forthcoming). At the same time, the proving the quality of education or public health. F I G U R E 5.1 6 P E R C E P T I O N O F I N C O M E TA X 100 80 Percent of respondents 60 40 20 0 Reduce more Reduce No need to do anything Increase Increase more Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the current situation, how far do you want the income tax collection to be increased? F I G U R E 5.1 7 P E R C E P T I O N O F I N C O M E TA X R AT E S A M O N G THE RICH BY ACTUAL INCOME 100 80 Percent of respondents 60 40 20 0 Bottom quintile Top quintile Lower The same Higher Much higher Source: DOSM-World Bank Survey of Perceptions of Inequality and Intergenerational Mobility (2023). Note: Based on the current situation, at what rates should those with higher incomes pay income taxes relative to those with lower incomes? Those with higher incomes should pay taxes at a rate that is (much higher, higher, the same, lower, much lower). 95 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 5.5 BROADENING THE INEQUALIT Y MONITORING TOOLKIT T racking inequality trends underestimated because of missing incomes at is the first essential step to the top of the distribution. address them. Malaysia has a long and strong tradition in Malaysia can track how all its people are far- measuring welfare and its dis- ing toward the country’s aim of becoming an tribution. The Gini is perhaps HIC. With this idea of improving the tracking the most well-known measure of progress on equity, the World Bank has intro- to follow inequality trends, and this headline duced the prosperity gap (PG) measure (World number can focus public and policy attention Bank. 2024c).74 At the global level, this is the on this important issue. For instance, the World shortfall from a prosperity standard of US$25 Bank’s new corporate score card includes track- per day. Malaysia can use this to benchmark it- ing high-Gini countries in order to track progress self against progress in other countries. Accord- on inequality.73 The Gini has some limitations, for ing to this measure (Global PG in Figure 5.18), in example, its emphasis on the middle of the distri- 2019, the PG stood at 1.3 while the EAP average bution or the lack of subgroup decomposability; was 3.3. This means Malaysians’ average shortfall additional measures of inequality could hence be from US$25 stands well below that in the EAP re- introduced as complements. The Department of gion, although the region has been catching up. Statistics Malaysia (DOSM) has recently expand- In addition, this report has presented an adapta- ed the inequality measures it reports; the Theil tion of this measure to Malaysia using a higher and Atkinson indices, as well as the Gini, have standard than US$25, given by the household per been included. Another useful measure, which is capita equivalent to the income needed to reach both simple to construct and communicate, and high-income status. This helps to focus on how which the DOSM is already reporting in reports Malaysians are faring toward the country’s goal of of the Household Income, Expenditure and Ba- becoming an HIC. sic Amenities Survey (HIES/BA) survey, is the share of national income going to each quintile Inequality of market income can be low- or decile of the distribution. It shows how the in- ered through taxes and transfers. This report comes in one place of the distribution—the top emphasized that inequality reduction could be deciles, the middle, and the bottom ones—have quickly achieved through the fiscal system; how- grown relative to others. For example, a decrease ever, the degree to which this is achieved varies in the Gini could be due to the middle catching considerably across countries and over time. In up to the top while the bottom is lagging, or due Malaysia, the level of inequality is only about 3 to the top falling rapidly. Administrative tax re- points lower when considering taxes and (cash) cords can complement this analysis since income transfers than when looking at market incomes data as captured by household surveys are likely alone, and this has barely changed over the past F I G U R E 5 .1 8 20 PROSPERIT Y GAP I N M A L AY S I A A N D 15 E AP (1990–2019) Prosperity gap Global PG EAP Source: World Bank (2024c) 10 Malaysian PG and World Bank staff calcula- tions based on DOSM HIES/BA. Global PG Malaysia Note: Global prosperity gap (z = 5 US$25 per person/day). Malay- sian prosperity gap (z = approx- imately US$36 per person/day). 0 EAP = East Asia and Pacific; PG 1990 1997 2004 2011 2018 = prosperity gap. 73. Gini for Malaysia, measured using the World Bank’s international welfare aggregate standardization method for global reporting, would classify the country as borderline “high inequality,” which corresponds to a Gini of 40 or more. The current Gini of 39, measured using DOSM’s methodology, would be just below this threshold. 74. See World Bank (2024c) and https://documentsinternal.worldbank.org/search/34213739 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 96 two decades (figure 2.3). Globally, SDG 10.4.2 geolocated data and modeling climate hazards at (the redistributive impact of fiscal policy)75 has a small local level. been added to the battery of indicators to sys- tematically track country-level progress in the Better and more systematic measurement of contribution of fiscal policy to more equitable service quality would help improve service societies. Tracking how much changes in fiscal delivery. Despite considerable advances in im- policy reduce inequality from market to post-fis- proving access to basic services and increasing cal incomes is also a useful benchmark to assess educational attainment, there remain gaps in the redistributive impact of a specific policy that opportunities that reflect that service quality is is being considered for implementation. an increasingly important source of persistent in- equality. Stunting rates remain high despite high Mobility and inequality of opportunity access to antenatal services and skilled deliveries; should be tracked alongside inequality. Be- learning poverty rates are high despite universal ing able to track mobility trends and inequality access to primary education; and returns on high- of opportunity is essential for middle-income er education are falling but are still dispropor- countries (World Bank 2024b). High inequality of tionally higher for workers from richer families. market income is not necessarily a negative in the Understanding and addressing these gaps more context of high mobility, where effort is closely effectively requires further information regarding linked to outcomes and everyone has access to service quality and the subnational gaps that need an expanded set of opportunities that allow them focus. For example, internet access is close to to make choices. Measuring and tracking income universal but there is a lack of information on the mobility and inequality of opportunity, as well quality or affordability of connections. Standard as pre- and post-fiscal inequality, can shed light questions that can be added to household sur- into whether the level of inequality should be a veys can be easy to implement. For instance, the concern. Tracking inequality of opportunity re- World Health Organization and United Nations quires capturing in a single data source (such as Children’s Fund (WHO-UNICEF) Joint Moni- the HIES/BA) information that speaks about in- toring Programme76 has developed questions to dividuals’ fixed characteristics, which reflect their capture the quality, availability, and affordabili- predetermined lot in life (e.g., one’s gender and ty of water sanitation and hygiene services, and ethnicity), but also about the place of birth and the World Bank’s Living Standards Measurement parents’ education. This information can be used Study program has developed similar guidelines to construct an index like the Inequality of Op- for measuring household energy use (World portunity Index referenced earlier in the report, Bank and WHO 2021). In other areas, the quali- but also rigorously analyze the determinants of ty of public services could be understood better outcomes. In turn, panel survey data that follow using administrative data sets from line ministries the same households over time are essential for or public facilities. Finally, measuring the quality measuring income mobility, which can also be of public services also involves tracking spending, useful to better measure the impact of shocks. including at the subnational level, and rigorously Capturing the impact of shocks, particularly measuring the effectiveness of policy interventions. those related to climate risks, would also require F I G U R E 5.1 9 S TAT I S T I C A L Pillar 1: Data use DIM. 1.5 80 PERFORMANCE I N D I C AT O R S PILL ARS IN Pillar 2: Data services DIM. 2.1 DIM. 2.2 DIM. 2.4 88 M A L AY S I A Pillar 3: Data products DIM. 3.1 DIM. 3.2 DIM. 3.3 DIM. 3.4 83 Pillar 4: Data sources DIM. 4.1 DIM. 4.2 DIM. 4.3 DIM. 4.4 75 Source: World Bank SPI: https://www.worldbank. org/en/programs/statisti- cal- performance-indica- Pillar 5: Data infrastructure DIM. 5.2 55 tors/Framework. 75. The redistributive impact of a fiscal policy indicator is defined as the Gini index for prefiscal per capita (or equivalized) income less the Gini index for post-fiscal per capita (or equivalized) income. 76. https://washdata.org/monitoring/methods/core-questions. 97 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA A broad data agenda to support inequali- A forward-looking research agenda on in- ty measurement. Previous recommendations equality. Finally, there are a few areas of further require collecting and using additional data. A research that the report has uncovered as import- robust and advanced statistical system in Malay- ant to further understand inequality in Malaysia. sia is needed to support these needs. Efforts to The list is not exhaustive, but among them are (1) assess the performance of and improvements to collecting more detailed data on specific percep- statistical systems are tracked worldwide using tions, for example, to understand mobility expec- the World Bank’s Statistical Performance Indica- tations; (2) looking into the root causes of high tors framework. Malaysia performs well overall, in stunting in Malaysia and looking into other coun- the 4th quintile of global performance, but could try experiences in reducing stunting rates; (3) do more, especially in the “data infrastructure” exploring administrative data to shed light into component, which refers to both hard infrastruc- ethnicity-based gaps in education outcomes; (4) ture (legislation, governance, standards) and soft disentangling the various potential explanations infrastructure (skills, partnerships), as well as the for the smaller skill premium among poorer fam- financial resources to deliver data products and ilies; and (5) understanding the causes of limited services (Figure 5.19). productivity in agriculture in the lagging regions. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 98 REFERENCES Acemoglu, Daron, and David Autor. 2011. “Skills, Tasks, and Technologies: Implications for Employ- ment and Earnings.” In Handbook of Labor Economics, Volume 4, Part B, edited by David Card and Orley Ashenfelter, 1043–171. Elsevier. Almond, Douglas, and Janet Currie. 2011. “Killing Me Softly: The Fetal Origins Hypothesis.” Journal of Economic Perspectives 25 (3): 153–72. Alvaredo, Facundo. 2011. “A Note on the Relationship between Top Income Shares and the Gini Coef- ficient.” Economics Letters 110 (3): 274–77. Alvaredo, F., and J. Londoño-Velez. 2013. “High Incomes and Personal Taxation in a Developing Econ- omy: Colombia 1993–2010.” Commitment to Equity (CEQ) Working Paper 12, Center for Inter-American Policy and Research; Department of Economics, Tulane University; and In- ter-American Dialogue. Arampatzi, Efstratia, Martijn Burger, Elena Ianchovichina, Tina Röhricht, and Ruut Veenhoven. 2015. “Unhappy Development: Dissatisfaction with Life in the Wake of the Arab Spring.” Policy Re- search Working Paper 7488, World Bank, Washington, DC. Arias, Omar S., J. Humberto Lopez, William F. Maloney, Guillermo E. Perry, and Luis Serven. 2006. Poverty Reduction and Growth: Virtuous and Vicious Circles. World Bank Latin American and Carib- bean Studies. Washington, DC: World Bank. ADB (Asian Development Bank). 2023. Malaysia Ageing and Retirement Survey Wave 2 (2021–2022): Sur- vey Report. Manila: ADB. Atkinson, A. B., and T. Piketty. 2010. Top Incomes: A Global Perspective. Oxford: Oxford University Press. Auten, Gerald, and David Splinter. 2024. “Income Inequality in the United States: Using Tax Data to Measure Long-Term Trends.” Journal of Political Economy 132 (7): 2179–227. Bachas, Pierre, Lucie Gadenne, and Anders Jensen. 2023. “Informality, Consumption Taxes, and Redis- tribution.” The Review of Economic Studies 91 (5): 2604–34. Bárány, Zsófia L., and Christian Siegel. 2018. “Job Polarization and Structural Change.” American Eco- nomic Journal: Macroeconomics 10 (1): 57–89. Banerjee, Abhijit V., Rema Hanna, Gabriel E. Kreindler, and Benjamin A. Olken. 2017. “Debunking the Stereotype of the Lazy Welfare Recipient: Evidence from Cash Transfer Programs.” The World Bank Research Observer 32 (2): 155–84. Barros, Ricardo, Mirela Carvalho, Samuel Franco, and Rosane Mendonça. 2010. “Markets, the State, and the Dynamics of Inequality in Brazil.” In Declining Inequality in Latin America: A Decade of Progress? edited by L. F. Lopez-Calva and N. Lustig, 134–174. Washington, DC: Brookings Institution and United Nations Development Programme. Blumenstock, Joshua E., Gabriel Cadamuro, and Robert On. 2015. “Predicting Poverty and Wealth from Mobile Phone Metadata.” Science 350 (6264): 1073–76. Bussolo, Maurizio, and Luis F. Lopez-Calva. 2014. Shared Prosperity: Paving the Way in Europe and Central Asia. Washington, DC: World Bank. 99 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Chattopadhyay, Soumya, and Carol L. Graham. 2015. From Arab Spring to Fall: Some Insights from the Eco- nomics of Well-Being. Washington, DC: Brookings Institution. Chaudhuri, Shubham. 2003. “Assessing Vulnerability to Poverty: Concepts, Empirical Methods, and Illustrative Examples.” Department of Economics, Columbia University. Chetty, Raj, Nathaniel Hendren, and Lawrence Katz. 2016. “The Effects of Exposure to Better Neigh- borhoods on Children: New Evidence from the Moving to Opportunity Experiment.” American Economic Review 106 (4): 855–902. Chi, Guanghua, Han Fang, Sourav Chatterjee, and Joshua E. Blumenstock. 2022. “Microestimates of Wealth for All Low- and Middle-Income Countries.” Proceedings of the National Academy of Sci- ences 119 (3): e2113658119. Colgan, Brian. 2023. “EU-SILC and the Potential for Synthetic Panel Estimates.” Empirical Economics 64 (August): 1247–80. Credit Suisse. 2014. Global Wealth Report 2014. Zurich: Credit Suisse Research Institute. Dang, Hai-Anh, and Peter Lanjouw. 2017. “Welfare Dynamics Measurement: Two Definitions of a Vul- nerability Line and Their Empirical Application.” Review of Income and Wealth 63 (4): 633–60. Dang, Hai-Anh, and Peter Lanjouw. 2023. “Measuring Poverty Dynamics with Synthetic Panels Based on Repeated Cross Sections.” Oxford Bulletin of Economics and Statistics 85 (3): 599–622. Dang, Hai-Anh, Peter Lanjouw, Jill Luoto, and David McKenzie. 2014. “Using Repeated Cross-Sections to Explore Movements into and out of Poverty.” Journal of Development Economics 107 (March): 112–28. Deaton, Angus, and Christina Paxson. 1995. “Saving, Inequality and Aging: An East Asian Perspective.” Asia-Pacific Economic Review 1 (1): 7–19. Doan, Miki K., Ruth Hill, Stephane Hallegatte, Paul Corral, Ben Brunckhorst, Minh Nguyen, Samuel Freije-Rodriguez, and Esther Naikal. 2023. “Counting People Exposed to, Vulnerable to, or at High Risk from Climate Shocks.” Policy Research Working Paper 10619, World Bank, Washing- ton, DC. Economics Malaysia. 2023. “Targeting on Household Income Doesn’t Make Sense and Has to Go.” Economics Malaysia, January 25, 2023. http://econsmalaysia.blogspot.com/2023/01/target- ing-on-household-income-doesnt.html. European Commission. 2017. My Region, My Europe, Our Future: Seventh Report on Economic, Social and Territorial Cohesion. Brussels: European Commission. Ferreira, Francisco H. G., Phillippe G. Leite, and Julie A. Litchfield. 2008. “The Rise and Fall of Brazilian Inequality: 1981–2004.” Macroeconomic Dynamics 12 (S2): 199–230. Ferreira, Francisco H. G., Sergio P. Firpo, and Julian Messina. 2017. “Ageing Poorly? Accounting for the Decline in Earnings Inequality in Brazil, 1995–2012.” Policy Research Working Paper 8018, World Bank, Washington, DC. Ferreira, Francisco H. G., Christoph Lakner, Mario A. Lugo, and Berk Özler. 2014. “Inequality of Oppor- tunity and Economic Growth: A Cross-Country Analysis.” Policy Research Working Paper 6915, World Bank, Washington, DC. Filmer, D and Pritchett, L.H. 2001. “Estimating wealth effects without expenditure data--or tears: an application to educational enrollments in states of India.” Demography 38 (1): 115-32. Firpo, Sergio, Nicole M. Fortin, and Thomas Lemieux. 2018. “Decomposing Wage Distributions Using A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 100 Recentered Influence Function Regressions.” Econometrics 6 (2): 1–40. Forbes. 2023a. “Malaysia’s 50 Richest.” https://www.forbes.com/malaysia-billionaires/. Forbes. 2023b. “World’s Billionaires List.” https://www.forbes.com/billionaires/. Gandhi, Aditi, and Michael Walton. 2012. “Where Do India’s Billionaires Get Their Wealth?” Economic and Political Weekly 47 (40): 10–14. Garcia, J. L., J. J. Heckman, D. E. Leaf, and M. J. Prados. 2017. “Quantifying the Life-Cycle Benefits of a Prototypcal Early-Childhood Program.” NBER Working Paper 23479, National Bureau of Eco- nomic Research, Cambridge, MA.  Gasparini, Leonardo, Sebastián Galiani, Guillermo Cruces, and Pablo Acosta. 2011. “Educational Up- grading and Returns to Skills in Latin America: Evidence from a Supply-Demand Framework, 1990–2010.” Policy Research Working Paper 5921, World Bank, Washington, DC. Ghorpade, Yashodhan, Amanina Binti Abdur Rahman, Alyssa Farha Binti Jasmin, Natalie Fang Ling Cheng, and Soonhwa Yi. 2024. Informal Employment in Malaysia: Trends, Challenges and Opportuni- ties for Reform. Washington, DC: World Bank. Goldin, C. 2021. Career and Family: Women’s Century-Long Journey toward Equity. Princeton, NJ: Princeton University Press. Grosh, M. Leite, P. Wai-Poi, M. and Tesliuc, E. 2022. Revisiting Targeting in Social Assistance: A New Look at Old Dilemmas. Human Development Perspectives. Washington, D.C.: World Bank. Gupta, Dipak. 1990. The Economics of Political Violence. New York: Praeger. Gutierrez, Catalina, Claudia Orecchia, Pierella Paci, and Pieter Maria Serneels. 2007. “Does Employ- ment Generation Really Matter for Poverty Reduction?” Policy Research Working Paper 4432, World Bank, Washington, DC. Hallegatte, Stephane, Marianne Fay, and Edward B. Barbier. 2018. “Poverty and Climate Change: Intro- duction.” Environment and Development Economics 23 (3): 217–33. doi:10.1017/S1355770X18000141. Hanewald, Katja, Ruo Jia, and Zining Liu. 2021. “Why Is Inequality Higher among the Old? Evidence from China.” China Economic Review 66 (April): 101592. https://doi.org/10.1016/j.chieco.2021.101592. Hasell, J. 2023. “Income Inequality Before and After Taxes: How Much Do Countries Redistribute Income?” Our World in Data, July 3, 2023. https://ourworldindata.org/income-inequality-be- fore-and-after-taxes. Heckman, James J. 2008. “Schools, Skills, and Synapses.” Economic Inquiry 46 (3): 289–324. https://doi. org/10.1111/j.1465-7295.2008.00163.x. Heckman, James J., S. H. Moon, R. Pinto, P. Savelyev, and A. Yavitz. 2010. “The Rate of Return to the High/Scope Perry Preschool Program.” Journal of Public Economics 94 (1–2): 114–28. Hill, R., T. Nguyen, and M. Khanh Doan. 2024. “Climate and Equity: A Framework to Understand Wel- fare Impacts and Guide Policy Action.” Prosperity Insight Series, World Bank, Washington, DC. Hoy, Christopher Alexander, Yeon Soo Kim, Minh Cong Nguyen, Mariano Ernesto Sosa, and Sailesh Tiwari. 2023. “Building Public Support for Reducing Fossil Fuel Subsidies: Evidence across 12 Middle-Income Countries.” Policy Research Working Paper 10615, World Bank, Washington, DC. Hungerford, Thomas. 2020. “The Course of Income Inequality as a Cohort Ages into Old-Age.” Journal of Economic Inequality 18 (August): 71–90. 101 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA Inchauste, G. forthcoming. IPCC (Intergovernmental Panel on Climate Change). 2021a. Sixth Assessment Report. Geneva, Switzer- land: IPCC. https://www.ipcc.ch/assessment-report/ar6/. IPCC. 2021b. “Summary for Policymakers.” In Climate Change 2021: The Physical Science Basis. Contri- bution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva, Switzerland: IPCC. https://www.ipcc.ch/report/ar6/wg1/downloads/report/ IPCC_AR6_WGI_SPM_final.pdf. IPH (Institute for Public Health). 2016. National Health & Morbidity Survey 2016 (NHMS 2016). Kuala Lumpur: Ministry of Health Malaysia. IPH. 2022. National Health & Morbidity Survey 2022 (NHMS 2022). Kuala Lumpur: Ministry of Health Malaysia. Jenkins, S. P., R. V. Burkhauser, S. Feng, and J. Larrimore. 2011. “Measuring Inequality Using Censored Data: A Multiple-Imputation Approach to Estimation and Inference.” Journal of the Royal Statis- tical Society: Series A (Statistics in Society) 174 (1): 63–81. Keefer, Philip, and Stephen Knack. 2002. “Polarization, Politics and Property Rights: Links Between Inequality and Growth.” Public Choice 111 (1–2): 127–54. Khalid, M., and Li Yang. 2021. “Income Inequality and Ethnic Cleavages in Malaysia: Evidence from Distributional National Accounts (1984–2014).” Journal of Asian Economics 72 (February): 101252. Kleven, Henrik, Camille Landais, and Gabriel Leite-Mariante. 2023. “The Child Penalty Atlas.” NBER Working Paper 31649, National Bureau of Economic Research, Cambridge, MA. Kleven, Henrik, Camille Landais, Johanna Posch, Andreas Steinhauer, and Josef Zweimüller. 2024. “Do Family Policies Reduce Gender Inequality? Evidence from 60 Years of Policy Experimenta- tion.” American Economic Journal: Economic Policy 16 (2): 110–49. Kraay, Aart. 2018. “Methodology for a World Bank Human Capital Index.” Policy Research Working Paper 8593, World Bank, Washington, DC. Kraay, Aart C., Christoph Lakner, Berk Ozler, Benoit Marie A. Decerf, Dean Mitchell Jolliffe, Olivier Christian Brigitte Sterck, and Nishant Yonzan. 2023. “A New Distribution Sensitive Index for Measuring Welfare, Poverty, and Inequality.” Policy Research Working Paper 10470, World Bank, Washington, DC. Levy, Santiago, and Luis-Felipe López-Calva. 2019. “Persistent Misallocation and the Returns to Educa- tion in Mexico.” Policy Research Working Paper 8690, World Bank, Washington, DC. López-Calva, Luis-Felipe, and Eduardo Ortiz-Juarez. 2014. “A Vulnerability Approach to the Definition of the Middle Class.” Journal of Economic Inequality 12 (January): 23–47. https://doi.org/10.1007/ s10888-012-9240-5. Lustig, Nora, Luis F. López-Calva, and Eduardo Ortiz-Juarez. 2013. “Declining Inequality in Latin Amer- ica in the 2000s: The Cases of Argentina, Brazil, and Mexico.” World Development 44 (April): 129–41. Muller, Andreas. 2008. “Clarifying Poverty Decomposition.” Working Papers in Economics 217, Depart- ment of Economics, Göteborg University, Göteborg, Sweden. Narayan, Ambar, Roy Van der Weide, Alexandru Cojocaru, Christoph Lakner, Silvia Redaelli, Daniel Gerszon Mahler, Rakesh Gupta, N. Ramasubbaiah, and Stefan Thewissen. 2018. Fair Progress? Economic Mobility across Generations around the World. Washington, DC: World Bank. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 102 OECD, Hanushek, E., Woessmann, L. 2015. Universal Basic Skills: What Countries Stand to Gain. Paris: OECD Publishing. Piketty, Thomas, and Emmanuel Saez. 2003. “Income Inequality in the United States, 1913–1998.” The Quarterly Journal of Economics 118 (1): 1–39. Piketty, Thomas, Emmanuel Saez, and Gabriel Zucman. 2018. “Distributional National Accounts: Meth- ods and Estimates for the United States.” The Quarterly Journal of Economics 133 (2): 553–609. https://doi.org/10.1093/qje/qjx043. Ravallion. 1998. Poverty Lines in Theory and Practice. LSMS Working Paper 133, World Bank, Washington, DC. Record,R.J., Larson, B.R., Teh Sharifuddin, S.B.; Chong, Y.K. 2018. Malaysia’s Digital Economy : A New Driver of Development. Washington, D.C.: World Bank Group. Rodríguez-Pose, Andrés, Federico Bartalucci, Nancy Lozano-Gracia, and Maria Eugenia Davalos. 2024. “Overcoming Left-Behindness: Moving Beyond the Efficiency versus Equity Debate in Territo- rial Development.” Policy Research Working Paper 10734, World Bank, Washington, DC. Rongen, Gerton, and Peter Lanjouw. 2024. The Only Way Is Up? Economic Mobility in Malaysia in the 21st Century. Policy Research Working Paper 10991, World Bank, Washington, DC. Rongen, Gerton, Zainab Ahmad, Peter Lanjouw, and Kenneth Simler. 2024. “Regional and Ethnic In- equalities in Malaysian Poverty Dynamics.” Journal of Economic Inequality 22 (1): 101–30. Roser, Max, and Hannah Ritchie. 2018. “Optimism and Pessimism.” Our World in Data, July 27, 2018. https://ourworldindata.org/optimism-and-pessimism. Shah, Syed Muzzamil Hussain, Zahiraniza Mustaffa, and Khamaruzaman Wan Yusof. 2017. “Disasters Worldwide and Floods in the Malaysian Region: A Brief Review.” Indian Journal of Science and Technology 10 (2): 1–9. Sharot, Tali. 2011. “The Optimism Bias.” Current Biology 21 (23): R941–45. Shorrocks, Anthony F. 2013. “Decomposition Procedures for Distributional Analysis: A Unified Frame- work Based on the Shapley Value.” Journal of Economic Inequality 11 (January): 99–126. https://doi. org/10.1007/s10888-011-9214-z. Skoufias, E., and J. Baez. 2021. “Operationalizing Vulnerability to Poverty.” Working paper, World Bank, Washington, DC. Sosa, Mariano, and Matthew Wai-Poi. Forthcoming. “Fiscal Policy and Equity in Developing Countries: A Survey of International Patterns and Lessons.” Background paper for Poverty and Shared Prosperity 2022, World Bank, Washington, DC. Stantcheva, Stefanie. 2021. “Understanding Tax Policy: How Do People Reason?” The Quarterly Journal of Economics 136 (4): 2309–69. https://doi.org/10.1093/qje/qjab033. Ting, H., T-Y. Lim, E. C. de Run, H. Koh, and M. Sahdan. 2018. “Are We Baby Boomers, Gen X and Gen Y? A Qualitative Inquiry into Generation Cohorts in Malaysia.” Kasetsart Journal of Social Sciences 39 (1): 109–15. UNDP (United Nations Development Programme). 2022. Human Development Report 2021/2022: Uncer- tain Times, Unsettled Lives: Shaping Our Future in a Transforming World. New York: UNDP. UNICEF (United Nations Children’s Fund). 2018. Children Without: A Study of Urban Child Poverty and Deprivation in Low-Cost Flats in Kuala Lumpur. Putrajaya, Malaysia: UNICEF. UNICEF (East Asia and the Pacific Regional Office) and the Centre for Justice and Crime Prevention. 103 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 2020. Our Lives Online: Use of Social Media by Children and Adolescents in East Asia—Opportunities, Risks and Harms. Bangkok: UNICEF. UNICEF, WHO (World Health Organization), and World Bank. 2023. “Joint Child Malnutrition Es- timates (JME)—Levels and Trends—2023 edition.” https://data.unicef.org/resources/jme-re- port-2023/. Van der Weide, Roy, Christoph Lakner, Daniel G. Mahler, Ambar Narayan, and Rakesh Ramasubbaiah. 2021. “Intergenerational Mobility around the World.” Policy Research Working Paper 9707, World Bank, Washington, DC. World Bank. 2005. World Development Report 2006: Equity and Development. Washington, DC: World Bank. World Bank. 2009a. Job Generation and Growth (JoGG) Decomposition Tool: Understanding the Sectoral Pat- tern of Growth and its Employment and Productivity Intensity. Reference Manual and User’s Guide Ver- sion 1.0. Poverty Reduction and Economic Management (PREM). Washington, DC: World Bank. World Bank. 2009b. World Development Report 2009: Reshaping Economic Geography. Washington, DC: World Bank. World Bank. 2016a. Poverty and Shared Prosperity: Taking on Inequality. Washington, DC: World Bank. World Bank. 2016b. Indonesia’s Rising Divide. Why Inequality Is Rising, Why It Matters and What Can Be Done. Washington, DC: World Bank. World Bank. 2019. “Inequality of Opportunity: New Measurements Reveal the Consequences of Un- equal Life Chances.” Feature story, March 28, 2019. https://www.worldbank.org/en/news/ feature/2019/03/28/inequality-of-opportunity-new-measurements-reveal-the-consequenc- es-of-unequal-life-chances. World Bank. 2020. 2019 Enterprise Survey. Washington, DC: World Bank. World Bank. 2021a. Aiming High: Navigating the Next Stage of Malaysia’s Development. Washington, DC: World Bank. World Bank. 2021b. Malaysia Economic Monitor: Staying Afloat. Washington, DC: World Bank. World Bank. 2022a. Malaysia Economic Monitor: Catching Up: Inclusive Recovery Growth for Lagging States. June 2022. Washington, DC: World Bank. World Bank. 2022b. “Malaysia: High-Frequency Monitoring of COVID-19 Impacts on Households 2021–2022, Rounds 1–3.” World Bank, Washington, DC. World Bank. 2022c. Poverty and Shared Prosperity 2022: Correcting Course. Washington, DC: World Bank. World Bank. 2022d. The State of Global Learning Poverty: 2022 Update. Washington, DC: World Bank. World Bank. 2022e. Philippines Country Climate and Development Report. Washington, DC: World Bank. World Bank. 2022f. Vietnam Country Climate and Development Report. Washington, DC: World Bank. World Bank. 2023a. Malaysia Economic Monitor—Malaysia’s Economic Outlook: A Review of Key Trends and Challenges. Washington, DC: World Bank. World Bank. 2023b. “World Bank Development Indicators 2023.” World Bank, Washington, DC World Bank. 2023c. Malaysia Economic Monitor: Raising the Tide, Lifting All Boats. Washington, DC: World Bank. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 104 World Bank. 2023d. Building Public Support for Energy Subsidy Reforms. What Will It Take? Washington, DC: World Bank. World Bank. 2023e. Services for Development. East Asia and the Pacific Economic Update, October 2023. Washington, DC: World Bank. World Bank. 2024a. Malaysia Economic Monitor: Bending Bamboo Shoots: Strengthening Foundational Skills. Washington, DC: World Bank. World Bank. 2024b. World Development Report 2024: The Middle-Income Trap. Washington, DC: World Bank. World Bank. 2024c. Poverty, Prosperity, and Planet Report 2024: Pathways out of the Polycrisis. Washing- ton, DC: World Bank. World Bank and BNM (Bank Negara Malaysia). 2024. Flood Risk Management: Leveraging Finance for Business Resilience in Malaysia. Washington, DC: World Bank. World Bank and WHO. 2021. Measuring Energy Access: A Guide to Collecting Data Using the Core Questions on Household Energy Use. LSMS Guidebook. Washington, DC: World Bank. Yonzan, N., B. Milanovic, S. Morelli, and J. Gornick. 2022. “Drawing a Line: Comparing the Estimation of Top Incomes Between Tax Data and Household Survey Data.” Journal of Economic Inequality 20 (April): 67–95. 105 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA TECHNICAL APPENDICES APPENDIX A . WELFARE AND INEQUALIT Y M E A S U R E M E N T I N M A L AY S I A Welfare measurement in Malaysia has a long tra- cial poverty reporting. It is also worth noting that dition. The first methodology to calculate the the published national poverty rates refer to the income poverty line was introduced in 1977 and proportion of households in poverty instead of the two major revisions have taken place in 2005 more conventional proportion of the population in and 2019. The welfare estimates in Malaysia are poverty. income based and derived from data from the household survey conducted twice in every five While this methodology is used by DOSM in years by the Department of Statistics Malaysia their official reporting of poverty, it does not (DOSM). The methodology for welfare measure- seem to be consistently used in other areas. First, ment and the poverty line in particular, is based unlike poverty, the official published Gini index on the cost of basic needs approach. In other uses total household income as the welfare aggre- words, the minimum requirement of nutrition- gate, without normalization for household size or al and nonfood items that are needed for each composition. As in poverty, inequality is calculat- of the household members to live a healthy and ed across households and not across individuals. active life in society. While the approach has re- The resulting Gini index estimates from DOSM mained the same across revisions, the specific se- are 2.0 to 2.5 percentage points lower than those lection of food and nonfood items included in the calculated by the more common approach that basic basket of goods was adapted to fit current uses household income per capita and popu- needs and to refine the methodology according lation weights. In addition, it appears that the to the latest recommendations. For instance, the welfare measure that is used to target social as- 2019 revision changed the reference population sistance is also total household income (see, for group for the poverty line determination to be example, Economics Malaysia 2023). This choice the bottom 20 percent of households, in order to distorts the profiles of the poor and vulnerable reflect the recommendation of Ravallion (1998) and leads to mistargeting social assistance, for ex- to focus on the consumption baskets of low-in- ample, away from larger but poorer families. come households. The average national poverty line more than doubled with the 2019 revision, While it is beyond the scope of the analysis in leading to an increase in the reported poverty this report to revise the poverty measurement incidence. methodology (i.e., decisions for the definition of the poverty line), the way welfare is measured has There is no single poverty line in Malaysia. The implications for the measures of inequality. In this poverty line methodology accounts for differenc- report, a consistent welfare aggregate that is in es in prices in the state and rural-urban locations. line with international practices, and which may It also takes into account the household composi- be different from the official reported statistics, tion by using an equivalence scale to account for will be applied. That is, using per capita house- different recommended nutrition intakes by age hold income instead of total household income and gender (although detailed information on the and reporting statistics that refer to population equivalence scales is not public). Combined, this rather than households (e.g., the percentage of results in tens of thousands of household-specific people in poverty, Gini across individuals). poverty lines that are used by DOSM in their offi- A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 106 Not accounting for household size or composi- er but poorer families higher in the distribution. tion results in 2–3 percentage points lower in- Not accounting for price differences within Ma- equality than when using household income per laysia (across states) would result in about 1 per- capita. Using total household income can distort centage point higher inequality estimates and the profile of households across the distribu- distorts the profile of households. tion, for example, it can lead to misplacing larg- F I G U R E A .1 G I N I I N D E X , 2 0 0 4 – 2 2 ( P E R C A P I TA V S . H O U S E H O L D T O TA L I N C O M E ) 46 44 42 40 38 36 2004 2007 2009 2012 2014 2016 2019 2022 Per capita household income Household income Source: World Bank staff calculations from DOSM HIES/BA. FIGURE A.2 G I N I I N D E X , 2 0 0 4 – 2 2 ( W I T H A N D W I T H O U T S PAT I A L D E F L AT I O N ) 49 47 45 43 41 39 37 35 2004 2006 2008 2010 2012 2014 2016 2018 2022 No spatial deflation Spatial deflated Source: World Bank staff calculations from DOSM HIES/BA. 107 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA A P P E N D I X B . S U R V E Y O F M A L AY S I A N S ON PERCEPTIONS OF INEQUALIT Y AND I N T E R G E N E R A T I O N A L M O B I L I T Y, 2 0 2 3 The Survey of Malaysians on Perceptions of In- views over a one-month period from March 10, equality and Intergenerational Mobility 2023 2023, to April 6, 2023. was undertaken specifically for the inequality study by the Department of Statistics Malaysia, in This survey aims to gauge public opinions toward close collaboration with the World Bank. A total inequality and mobility in Malaysia and if there is of 10,632 Malaysian households were randomly general support for a policy response. The survey selected from the 2022 National Household In- sample was designed to be nationally representa- come and Basic Amenities Survey, representing tive, with key demographic composition reflect- 7.9 million households throughout the states. The ing the population distribution in the 2020 Popu- respondents were heads of households. The data lation and Housing Census. Table B.1 outlines the collection was conducted via face-to-face inter- sample distribution of the survey. TA B L E B .1 S A M P L E D I S T R I B U T I O N O F S U R V E Y O F M A L AY S I A N S O N P E R C E P T I O N S O F I N E Q U A L I T Y A N D I N T E R G E N E R AT I O N A L MOBILIT Y 2023 Category Description Percentage of respondents (%) State Selangor 22 Johor 12 Perak 8 Sabah 8 Sarawak 8 FT Kuala Lumpur 7 Kedah 7 Penang 6 Kelantan 5 Pahang 5 Negeri Sembilan 4 Terengganu 3 Melaka 3 Perlis 1 FT Putrajaya 0 FT Labuan 0 Ethnicity Bumiputera 66 Chinese 25 Indian 8 Others 2 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 108 Age groups* Baby boomers (64 and above) 14 Gen X (43–63 years) 45 Gen Y (32–42 years) 28 Gen Z (18–31 years) 13 Highest Sijil Pelajaran Malaysia and below 64 education obtained Bachelor’s degree or master’s 18 Certificate or diploma 18 PhD 0 Status Private employee 52 activity Civil servant 12 Self-employed (registered) 11 Self-employed (not registered) 10 Elderly (non-retiree) 5 Retiree (civil servant) 5 Employer 2 Retiree (private employee) 2 Housewife 1 Others 0 Unemployed 0 Student 0 Unpaid family worker 0 Occupational groups Skilled 35 Semiskilled 42 Low skilled 8 Army 1 Others (unsure and others) 13 Source: Ting et al. 2018. Note: The age group follows categorization as proposed by Ting et al. (2018) to reflect the local context. 109 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA APPENDIX C. ADDITIONAL FIGURES AND TA BL E S F OR M O BIL I T Y E S T IM AT E S F I G U R E C .1 POVERT Y TR C H R O N I C Population O N S ( P Otransitions A N S I T Ipoverty shares P U L AT I O N S H A R E S ) Chronic poverty Non-poor in both rounds .4 .9 .3 .8 .7 .2 .6 .1 .5 0 .4 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Moving out of poverty Moving into poverty .4 .4 .3 .3 .2 .2 .1 .1 0 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Shares are based on synthetic panel estimates for household heads aged between 25 and 60 in the first round of each interval. The poverty line used is RM 527; note that the vertical axis scales vary across graphs Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds of shares for chronic poor. Poverty line at RM 527. FIGURE C.2 CHRONIC POVERT Y BY E THNICIT Y AND GEOGRAPHY Peninsular Malaysia East Malaysia Urban Peninsular Rural East Malaysia .7 .6 .5 .4 .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds, point estimates, and two standard deviation ranges in economically secure class (line of RM 1,590) in two consecutive survey waves. This group excludes the poor and vulnerable groups, based on the Chaudhuri (2003) and López -Calva and Ortiz-Juárez (2014) method and a 10 percent probability of falling into poverty to define the vulnerable. Poverty line at RM 527. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 110 FIGURE C.3 ABSOLUTE INCOME MOBILIT Y TRANSITIONS ( E C O N O M I C A L LY S E C U R E C L A S S ) Upward mobile Downward mobile Aspiring middle class in both Secure in both periods .6 .5 .4 .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Population shares are based on synthetic panel estimates for household heads aged between 25 and 60 in the first round of each interval. The poverty line is RM 920, the vulnerability lin RM 1590 Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds, point estimates, and two standard deviation ranges in economically secure class (line of RM 1,590) in two consecutive survey waves. This group excludes the poor and vulnerable groups, based on the Chaudhuri (2003) and López -Calva and Ortiz-Juárez (2014) method and a 10 percent probability of falling into poverty to define the vulnerable. F I G U R E C .4 R E L AT I V E M O B I L I T Y B E T W E E N I N C O M E Q U I N T I L E S , 2 0 0 4 – 0 7 T O 2 0 1 9 – 2 2 Upward mobile Downward mobile Quintile 1 in both Quintile 5 in both .4 .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Source: Rongen and Lanjouw 2024. Note: Upper and lower bounds, point estimates, and two standard deviation ranges. 111 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA F I G U R E C .5 R E L AT I V E M O B I L I T Y B E T W E E N I N C O M E Q U I N T I L E S , 2 0 0 4 – 0 7 T O 2 0 1 9 – 2 2 Quintile 1 in both Bumiputera Chinese Malaysian Indian Malaysian Share within the subgroup .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Quintile 5 in both Bumiputera Chinese Malaysian Indian Malaysian Share within the subgroup .3 .2 .1 0 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 2004-07 2007-09 2009-12 2012-14 2014-16 2016-19 2019-22 Bounds Point estimate +/- two standard deviations Population shares are based on synthetic panel estimates for household heads aged between 25 and 60 in the first round of each interval. Quintile groups are determined at population level Source: Rongen and Lanjouw 2024. Note: Quintiles are determined at the population level. TA B L E C .1 C O U N T E R FA C T U A L D ATA Panel 1: Parameters for counterfactual analysis Age Share of income recipients (%) Mean income Within group group (RM) MLD 2022 2043 2056 15–24 8.7 6.5 5.7 2,141 0.226 25–49 64.6 53.7 48.7 4,611 0.260 50–64 18.7 25.7 25.1 5,103 0.382 65+ 8.0 14.1 20.6 3,045 0.392 Total 100.0 100.0 100.0 4,362 0.291 Note: Shares give the projected proportion of income recipients (15 years and above) who are in the indicated age group. Mean income is for this sample of income recipients for the year 2022; it is gross monthly income received by the individual, expressed in 2016 Malay- sian ringgit (RM) and adjusted for spatial price differences. MLD = mean log deviation. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 112 Panel 2: Counterfactual inequality results MLD Within Between Total % change 2022 0.291 0.026 0.316 . 2043 0.308 0.025 0.333 5.4 2056 0.316 0.027 0.343 8.5     Percentage (%) Within Between Total   2022 91.9% 8.1% 100.0%   2043 92.4% 7.6% 100.0%   2056 92.1% 7.9% 100.0%   Source: Calculations based on UN World Population Prospects (medium variant) and DOSM HIES/BA data. Note: This panel presents the counterfactual results of how inequality in 2022 would have looked if population shares had been as pro- jected in the UN World Population Prospects for the years 2022, 2043, and 2056. Shares have been rescaled to go from full population to income recipients. The measure is the mean log deviation (MLD), decomposed in within- and between-group components. Additional figures within cohort inequality: GE(1), GE(-1), and Gini index F I G U R E C .6 I N E Q U A L I T Y W I T H I N A G E C O H O R T S — G E 1 ( S T R AT U M : U R B ), INDIVIDUAL INCOME Inequality within age cohorts - ge1 (stratum: urb) Individual income .6 .4 .2 0 2005 2010 2015 2020 2005 2010 2015 2020 2005 2010 2015 2020 Survey year Survey year Survey year 1945 1950 1955 1960 1965 1970 1975 1980 1985 The figure shows the development of inequality within each five-year birth cohort over time. 113 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE C. 7 INEQUALIT Y WITHIN AGE COHORTS—GE (-1) ( S T R AT U M : U R B ), I N C O M E R E C I P I E N T S Inequality within age cohorts - gem1 (stratum: urb) Individual recipients 2 1.5 1 .5 0 2005 2010 2015 2020 2005 2010 2015 2020 2005 2010 2015 2020 Survey year Survey year Survey year 1945 1950 1955 1960 1965 1970 1975 1980 1985 The figure shows the development of inequality within each five-year birth cohort over time. FIGURE C.8 INEQUALIT Y WITHIN AGE COHORTS—GINI ( S T R AT U M : U R B ), I N D I V I D U A L I N C O M E Inequality within age cohorts - gini (stratum: urb) Individual income .6 .5 .4 .3 .2 2005 2010 2015 2020 2005 2010 2015 2020 2005 2010 2015 2020 Survey year Survey year Survey year 1945 1950 1955 1960 1965 1970 1975 1980 1985 The figure shows the development of inequality within each five-year birth cohort over time. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 114 APPENDIX D. DEFINITIONS AND ME T HOD OL O GY F OR CL IM AT E R ISK A N A LY S I S DISASTER RISK Disaster risk measures the probability of a neg- ple and their assets which may be threatened by ative impact caused by a natural hazard. The natural hazards.78 While people (and their assets) United Nations Office for Disaster Risk Reduc- may get exposed, they may not be adversely im- tion (UNDRR) defines natural hazard as potential pacted by hazards if they are not vulnerable.79 To- events or trends that may cause loss of life, injury, gether, hazard (H), exposure (E), and vulnerability or other health impacts; and damage and losses to (V) drive disaster risk (R). More details about the property, infrastructure, livelihoods, service pro- methodology are available online.80 vision, ecosystems, and environmental resourc- es.77 Exposure characterizes the location of peo- F I G U R E D .1 H A Z A R D , R I S K , V U L N E R A B I L I T Y, A N D E X P O S U R E Hazard Risk Vulnerability Exposure Occurrence Occurrence Physical Location Size Location Socio-economic Type Intesity Type Size Size 77. https://www.undrr.org/terminology/hazard. 78. https://www.undrr.org/terminology/exposure. 79. https://www.undrr.org/terminology/vulnerability. 80. https://gfdrr.github.io/CCDR-tools/home.html. 115 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA HAZARD Natural events (including extreme events and b. Probabilistic, in the form of multiple long-term phenomena) are only termed hazards geodata layers, each representing a range of when they have the potential to harm people or hazard physical intensities (e.g., water depth cause property damage or social and economic [meters], wind speed [kilometers per hour]) disruption. The location of natural hazards pri- corresponding to a specific occurrence marily depends on geography, environmental frequency, measured as return period (RP), conditions (such as presence of water bodies, in years. This is the case for river and coastal slopes, vegetation), and natural processes, includ- floods. ing the influence of weather systems and tectonic Note that hazard models carry limitations relat- movements. Anthropic processes such as urban- ed to their applicability, as their quality depends ization, environmental degradation, and climate on scale, resolution, model, and input data. As a change can also influence the location, frequency rule of thumb, their fitness for application in the of occurrence, and intensity of natural hazards. context of a risk screening or assessment exercise These are known as risk drivers. depends on the scale of the risk analysis: local- ly sourced models are expected to be best fitted Hazard intensity can be modelled following two for local-scale assessment (e.g., city level), while different approaches: global models are best suited for national or sub- national estimates. In the context of developing a. Deterministic, in the form of an individual countries, however, a global model is often the geodata layer measuring the mean, median, only available source. In such cases, the applica- or maximum intensity of a hazard aggregating tion of the global model must be taken with cau- historical data and modeling. This is the case tion and correctly interpreted acknowledging the for droughts and air pollution. limitations. EXPOSURE Exposure describes the location of people and egories used as main indicators of risk, listed in assets that are prone to suffer an impact from figure D.2. natural hazards. We consider three exposure cat- FIGURE D.2 E X P O S U R E C AT E G O R I E S C O N S I D E R E D I N T H E A N A LY S I S A N D R E L AT E D I M PA C T T Y P E S Population • Potential impacts on health and mortality Built-up environment • Physical damage to buildings and infrastructural assets Agriculture and natural environment • Affected cropland and pastures • Crop production and livestock numbers A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 116 Each indicator is quantified by a specific metric: land environment are measured in terms of area population is described in terms of total count per (hectares) and share over total land area within area and as share of total population within an the administrative unit. administrative unit, while built-up and agricultural VULNERABILITY Vulnerability is determined by physical, social, event. These are measured using spatial indices economic, and environmental factors or process- based on demographics (sex, age composition, es, which increase the susceptibility of an individ- dependency rate) and socioeconomic statistics ual, their assets, or a community, to hazards [UN- (wealth, gross domestic product, and average DRR]. Two main components of vulnerability are salary, among others), and are semiquantitative typically accounted for in this note: metrics (index score; ranking). • Impact models, draw the relationship between the intensity of the hazard and the Not all exposure categories are affected in the predisposition of damage suffered by specific same way by physical hazards—some hazards exposed categories into actual impact; for are more relevant for one category than another. example, a flood depth of 0.5 meters is expected The impact model needs to be aligned with the to cause a low degree of impact in terms of hazard intensity metric, with the exposure cate- population mortality, while a 3 meters flood gory and with the socioeconomic conditions to would cause severe impacts. Impact models which they are applied to. For this reason, the can be quantitative, providing an absolute or availability of such models dictates the possible relative estimate of the damage (i.e., in terms combinations of hazard and exposure categories. of US dollars or percentage of total value); or Our approach relies on global- or regional-level qualitative, classifying the impact in nominal vulnerability models. Table D.1 identifies which categories. combinations are sustained by currently available • Socioeconomic conditions. Describes impact models, and which can only be classified the differential susceptibility of exposed in terms of exposure to hazard classes, defined categories to suffer damage, that is, areas under using hazard intensity thresholds based on litera- poverty conditions and high dependency rate ture studies. More details about the impact mod- are more likely to suffer damage compared to els for each combination are given in appendix C. wealthy communities, under the same hazard TA B L E D .1 AVA I L A B L E H A Z A R D , E X P O S U R E , A N D V U L N E R A B I L I T Y C O M P O N E N T S F R O M G L O B A L D ATA S E T S Exposure categories Agricultural Built-up assets​  Population  land  Hazard types [Physical [Mortality]​  [Production damage]​ losses]  River and coastal floods Exposure Probabilistic Impact model ​ Impact model​ by hazard [Water extent and depth] classes Landslides Exposure Exposure Deterministic by hazard by hazard [Landslide hazard index] classes classes Heat stress Exposure Probabilistic by hazard [Heat index] classes The socioeconomic component is added after the baseline risk calculation to highlight where the risk is expected to translate into the most severe impacts. 117 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA BA SEL INE RISK— E X PEC TED ANNUAL E X POSURE AND IMPAC T Baseline risk refers to an estimate related to the es that of the selected exposure layers. The im- historical period to which the data refer, as op- pact model translates the physical intensity unit posed to future risk (climate outlook). Baseline risk of a specific hazard into a damage factor (0 to 1), is calculated by combining geospatial hazard and which is then multiplied by the exposure layer to exposure data. Locations with no hazard or no obtain the impacted share over the total exposed exposure are excluded from calculations. The value (figure D.3). unit of analysis is set to a resolution that match- FIGURE D.3 E X A M P L E I M PA C T M O D E L Note: Example of hazard, exposure, and vulnerability components combined in a GIS environment. The flood hazard layer (blue) de- scribing water extent and depth (m) overlays the exposure layer (orange), which describes population count or built-up area. Where they match, there is an impact (pink), which is calculated as the product of the total exposure and the damage factor, driven by the impact model: depth-mortality function in the case of population (top-left box), depth-damage function in the case of built-up area (bottom-right box). ha = hectare; m = meter; RP = return period. When probabilistic hazard scenarios are avail- RPi+1), and then summing up to one value; else, if able, the expected annual impact (EAI) is calcu- no impact function is available, the expected an- lated by multiplying the impact from each event nual exposure (EAE) is calculated for a selected scenario with its exceedance probability (1/RPi–1/ hazard threshold. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 118 F I G U R E D .4 C O M P U TAT I O N O F A N N U A L E X P E C T E D I M PA C T O F N AT U R A L H A Z A R D S I N G E O S PAT I A L A N A LY T I C S Expected Annual Impact = EAI1 + EAI2 + EAI3 ... + EAIn Impact, I EAIn EAI6 EAI3 EAI5 EAI4 EAI2 EAI1 Annual Exceedance Probability, AEP (%) Note: The exceedance frequency curve highlights the relationship between the return period of each hazard and the estimated impact. The area below the curve represents the total annual damage considering all scenario probabilities. EAI = expected annual impact. Relative living standard maps are produced based variate maps.81 The bivariate maps provide ranks on the Relative Wealth Index (RWI) 2020. RWI explained by a 3x3 matrix, resulting in nine pos- values are weighted using the 2020 population to sible scores ranging from low risk/low poverty make the index demographically representative to high risk/high poverty. The matrix is built by (Filmer and Pritchett 2001) before calculating classifying poverty indicators into three quantiles the average for subnational administrative units and dividing risk indicators (EAE or EAI) of each (ADM2 level). These are combined with the risk hazard type into classes, according to thresholds maps (EAE or EAI) at the same level to obtain bi- shown in table D.2. TA B L E D . 2 S E L E C T E D R I S K C L A S S I F I C AT I O N APPROACH BY HA Z ARD Risk classification Risk indicator Unit Low Medium High Relative wealth index [-] -2–0.25 -0.25–0.25 >0.25 River flood x population EAI/km2 0–0.1 0.1–1 >1 Coastal flood x population EAI/km 2 0–0.01 0.01–0.05 >0.05 Landslide x population EAE/km2 0–25 25–250 >250 Heat stress​x population EAE/km2 1–100 100–300 >300 Note: EAE = expected annual exposure; EAI = expected annual impact; km2 = square kilometer. 81. The metrics for EAE/EAI are computed relative to ADM area (square kilometers) in order to normalize the values distribution, to avoid outliers that would cause an unrepresentative quantile splitting. 119 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA FIGURE D.5 R E L AT I V E W E A LT H I N D E X ( D I S T R I C T L E V E L M E A N ) Source: World Bank staff calculations from http://www.povertymaps.net/. Note: Relative wealth index are microestimates of the relative standard of living within countries produced using deidentified connec- tivity data, satellite imagery, and other nontraditional data sources (see: Blumenstock, Cadamuro, and On 2015; Chi et al. 2022). km = kilometer. FIGURE D.6 P O P U L AT I O N E X P O S U R E T O R I V E R F L O O D S I N R E L AT I O N T O L I V I N G S TA N D A R D S ( R W I ) Source: World Bank staff calculations from http://www.povertymaps.net/. Note: RWI = Relative Wealth Index; km2 = square kilometer. FIGURE D. 7 P O P U L AT I O N E X P O S U R E T O C O A S TA L F L O O D S I N R E L AT I O N T O L I V I N G S TA N D A R D S ( R W I ) Source: World Bank staff calculations from http://www.povertymaps.net/. Note: RWI = Relative Wealth Index; km2 = square kilometer. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 120 FIGURE D.8 P O P U L AT I O N E X P O S U R E T O L A N D S L I D E S I N R E L AT I O N T O L I V I N G S TA N D A R D S ( R W I ) Source: World Bank staff calculations from http://www.povertymaps.net/. Note: RWI = Relative Wealth Index; km2 = square kilometer. F I G U R E D .9 P O P U L AT I O N E X P O S U R E T O H E A D S T R E S S I N R E L AT I O N T O L I V I N G S TA N D A R D S ( R W I ) Source: World Bank staff calculations from http://www.povertymaps.net/. Note: RWI = Relative Wealth Index; km2 = square kilometer. CL IM AT E OU T L OOK The forward-looking analysis uses future climate referred to as Coupled Model Intercomparison projections to explore how environmental risks Projects (CMIP). The analysis relies on CMIP6 could develop spatially across the region. The data for modeling into the future, and takes into long-term averages of climate indices (observed account four climate change scenarios, referred or simulated) serve as the baseline conditions, to as Shared Socioeconomic Pathways (SSPs) in against which the effects of climate change CMIP6. These pathways cover the range of pos- are measured for future scenarios. Changes in sible future scenarios of anthropogenic drivers of projected climate indices against this baseline climate change by accounting for various future (anomalies) are used to estimate changes in nat- greenhouse gas emission trajectories, as well as a ural hazard frequency and intensity. Data from specific focus on carbon dioxide concentration climate models released under the Intergov- trajectories (IPCC 2021b). Three scenarios are ernmental Panel on Climate Change Sixth As- included in this analysis: sessment Report (AR) framework (IPCC 2021a) • SSP1-1.9. Emissions peak between 2040 and are used to establish estimates of baseline and 2060, declining by 2100. This results in 3-3.5°C future projected climate anomalies. ARs are sup- of warming by 2100. ported by coordinated climate modeling efforts • SSP2-4.5. Emissions continue to increase 121 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA through the end of the century, with resulting Each climate scenario predicts different spatial warming of 3.8–4.2°C. patterns, resulting in a range of possible futures • SSP3-7.0. Models describe a large emission in terms of intensities and frequencies of natural variability for this scenario. Warming in 2100 is hazards. Key climate variables connected to the estimated at 3.9–4.6 °C. changing patterns of precipitation and tempera- ture are collected from the World Bank Climate Knowledge Portal82 and summarized in table D.3. TA B L E D . 3 Unit of Hazard​ Associated climate indices​ C L I M AT E measurement VA R I A B L E S Rainfall >10 millimeters Days per year U N D E R LY I N G C L I M AT E Consecutive wet days Days per year Inland floods and PROJECTIONS landslides​ Maximum five-day millimeters precipitation Extremely wet days millimeters Coastal floods Sea level rise meters Days per year >23°C Note: WBGT = Heat stress​ WBGT heat index Wet Bulb Globe Days per year >30°C Temperature. F I G U R E D .1 0 1.4 Upper Secondary RETURNS TO 1.3 1.2 E D U C AT I O N B Y 1.1 QUANTILES OF THE 1 L ABOR INCOME .9 DISTRIBUTION, .8 2004–22 .7 .6 .5 .4 .3 .2 .1 0 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 Quantile 2004 2009 1.4 Vocational & above2014 2019 2007 2012 2016 2022 1.3 1.2 1.1 Source: : World Bank staff calculations 1 based on data from the World Bank Pov- erty and Inequality Platform (https://pip. .9 worldbank.org/). .8 .7 Note: Estimates represent the uncondi- tional quantile coefficients at different .6 percentiles of the employment income .5 distribution. The regressions control for .4 gender, potential experience, education .3 level, sector, ethnicity, strata, informality, .2 and occupation skill level. Since HIES/ .1 BA only has information on the level of 0 educational attainment, years of education, which are needed to estimate potential 0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1 experience, are proxied by the years Quantile required to complete each educational level. Alternative estimations using age and 2004 2009 2014 2019 age-squared yield very similar results. 2007 2012 2016 2022 82. https://climateknowledgeportal.worldbank.org/. A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA 122 123 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA CONNECT WITH US @Worldbankmalaysia @WB_Asiapasific bit.ly/WBMYblogs worldbank.org/malaysia | ifc.org | miga.org A FR E SH TA K E ON REDUCING INEQUALIT Y AND E N H A N C I N G M O B I L I T Y I N M A L AY S I A 125 A FRESH TAKE ON REDUCING INEQUALITY AND ENHANCING MOBILITY IN MALAYSIA