Financial Inclusion and Financial Capability in Morobe and Madang Provinces, Papua New Guinea An initial report of the Papua New Guinea National Financial Capability Survey Bank of Papua New Guinea Institute of National Affairs Bank of Papua New Guinea Papua New Guinea Institute for National Affairs The World Bank Financial Inclusion and Financial Capability in Morobe and Madang Provinces Papua New Guinea An initial report of the Papua New Guinea National Financial Capability Survey This Project is financially supported by the Korean Poverty Reduction and Socio-Economic Development Trust Fund II Cataloguing-in-Publication Data ISBN 9980-77-182-8 National Library Service—Papua New Guinea First published: June 2015 Published by: Institute of National Affairs P.O. Box 1530 Port Moresby NCD Papua New Guinea Copyright: This report is a joint product of the project team composed of staff and consultants from Bank of Papua New Guinea, the Institute of National Affairs and The World Bank. The findings, interpretations, and conclusions expressed in this report are entirely those of the authors and should do not necessarily reflect the views of Board of the Bank of Papua New Guinea, the Executive Directors of The World Bank or the governments they represent, or the Board of Institute of National Affairs. The Bank of Papua New Guinea, Institute of National Affairs, and The World Bank do not guarantee the accuracy of the data included in this work. The boundaries, colours, denominations, and other information shown on any map in this work do not imply any judgment on the part of Bank of Papua New Guinea or The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions: The material in this publication is subject to copyright. The Bank of Papua New Guinea, Institute of National Affairs and The World Bank encourage dissemination of this work for non-commercial purposes and will normally grant permission to reproduce portions of the work promptly as long as full attribution to this work is given. All other queries on rights and licenses, including subsidiary rights, should be addressed to Institute of National Affairs, P.O. Box 1530, Port Moresby, NCD, Papua New Guinea Note to Researchers: This report is an initial output from the study. The dataset is available to bona fide researchers. Data is held in SPSS and Excel versions. The code book is also available. Researchers wanting to use the dataset for analysis are requested to contact: Mr Boniface Aipi at BAipi@bankpng.gov. Disclaimer/Limitation: This report is a joint product of the project team composed of staff and consultants from Bank of Papua New Guinea, the Institute of National Affairs and the World Bank. The findings, interpretations, and conclusions expressed in this report are entirely those of the authors and do not necessarily reflect the views of the Board of the Bank of Papua New Guinea, the Executive Directors of The World Bank or the governments they represent or the Board of Institute of National Affairs. The Bank of Papua New Guinea, Institute of National Affairs, and The World Bank do not guarantee the accuracy of the data included in this work. The boundaries, colours, denominations, and other information shown on any map in this work do not imply any judgment on the part of Bank of Papua New Guinea or The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Authorship: This Report reflects the work of the Bank of Papua New Guinea, Institute of National Affairs and World Bank team comprising: Gae Kauzi, Bank of Papua New Guinea; Jonathan Sibley, World Bank consultant (principal author); Paul Barker, Institute for National Affairs; Wei Zhang, World Bank (Team Leader); John Gibson, World Bank consultant (sampling). Front cover: Artwork by Joe Scott of INA and photographs, all taken in Morobe and Madang Provinces, by Paul Barker. Foreword The Bank of Papua New Guinea and the World Bank are glad to present the initial findings from the first national survey of financial capacity among PNG population. Financial capability refers to the capacity to effectively manage financial sources over the life cycle and engage constructively with financial products and services. It is recognized as an essential skill for individuals in all walks of life and has become a public policy concern throughout the world from advanced economies to developing countries. It is a new challenge to individuals and households in low and middle- income economies where the traditional forms of family and community support are being replaced or blended gradually with formal social protection programs with innovative products and services. Traditional ways of doing business are also changing to cope with increasing competition at local, national and global levels. Understanding financial matters and making informed financial decisions among citizens will contribute to overall financial stability of a country as evidenced by the recent global financial crisis. Understanding the financial capability of PNG population is of particular relevance to policy makers, educators and researchers in PNG as we all know anecdotally that level of financial exclusion is high in PNG and the ability of many Papua New Guineans to manage money is limited as formal financial services are a rather new concept to the traditional communities in PNG. Increasing our knowledge and understanding is particularly important as the formal financial system continues to develop rapidly in PNG and households, whether urban or rural, are required to use an increasing number of financial instruments and to manage increasingly complex household finances. This National Financial Capability Survey aims to provide a detailed picture which can support policy and programmes to enhance financial inclusion and increase financial literacy of all Papua New Guineans. The initial findings from the surveys in Madang and Morobe provinces already provide a set of rich data that can be further analysed. Once completed the National Financial Capability study will provide a baseline measure of financial inclusion and financial capability for PNG. We would like to encourage sharing of key findings of this report among policy makers, researchers, educators and financial services providers so as to support the iii Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea development of more diversified, customized and affordable financial services and products to PNG population. Both the Government and BPNG are committed to increasing financial inclusion and financial literacy in PNG. The government is committed to expanding financial inclusion and enhancing financial literacy by incorporating the financial inclusion agenda in key 1 national plans as a development priority. The World Bank, a development partner with the Government of PNG, is committed to share the technical know-how and international best practices in financial inclusion program design and implementation. Loi Martin Bakani CMG, Governor, Bank of Papua New Guinea James Seward, Practice Manager, Finance and Markets Global Practice, The World Bank 1   Please refer to: “The Development Strategic Plan (2010–2030),” “The Medium Term Development Plan (2016–2017),” “Vision 2050”, “The National Informal Economy Policy (2011–2015).” iv Acknowledgements This report was prepared by a joint team of staff and consultants from The Bank of Papua New Guinea (BPNG), the Institute of National Affairs (INA), and The World Bank led by Gae Kauzi, Paul Barker and Ms. Wei Zhang respectively, with contributions from Elizabeth Genia, Ellison Pidik, Boniface Aipi and Augustine Birie of the Bank of Papua New Guinea, Mr. Emmanuel Peni of INA, Jonathan Sibley (Principal Author) and John Gibson (Sampling Advisor) of the World Bank. The project team appreciate the support and guidance provided by Governor Bakani of BPNG, Hormoz Aghdaey and James Seward, Practice Managers for East Asia and Pacific Region of the World Bank Finance and Markets Global Practice, in project conceptualization and implementation. The team is grateful to the peer reviewers of the report who provided valuable comments and suggestions from the peer reviewers: Chandana Kularatne, Siegfried Zottel, Dominic Sikakau and Rekha Reddy. The team also benefited tremendously from the technical advice and partnership from Jeff Liew, Senior Advisor of the Pacific Financial Inclusion Program (UNDP-PFIP). We owe our special thanks to the field survey management team in INA who took tremendous effort in managing data collection from some remote villages in Madang and Morobe provinces: Rufina Peter and Mary Maima, who were the initial project managers, Ivan Jemen, field coordinator for Morobe Provinece, Busa Jeremiah Wenogo, Senior Project Officer of Informal Economy, CIMC, who was the field supervisor for Madang province, Henry Yamo, Deputy Executive Officer, CIMC, Teine Korokoi—IT Officer, Maisy Talowani—Librarian & Assistant Logistics Coordinator, Allan Keneke, Kaupa Magis, Jenny Kaupa for their support to the field teams. Critical contribution was made by a team of hard-working local enumerators in Madang and Morobe provinces who carried out the households interviews: Killian Gemo, Jasmine Tasha Kong, Wiseng Umbingke, Lau Sorum Jr, Paul Naime, Jamilla Mare, Henry Basse, Jayson Namis, Melissa Poang, Russel Yakaua Ada, Zawepe Don, Estella Kawah, Morris Torokon, Mike Amel, Esther Wandil, Rose David, Florence Jesse, Sam Lance, Salome B. Gedisa, Gideon Vinguai, Lillian Dou, Jonathan Goga, Ignatiela Sala. The field survey teams were supported by the local police officers in the two provinces and we appreciate the time and effort by these officers who ensured the security of the field teams and facilitated the communications with the local communities: v Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Constable Micah Wrok, Constable Norberth Yuahahu, Constable Banas Kuder, Constable Bamea Zakang, and Constable Matthew Kepa. The World Bank security advisor Mr. Craig Stemp also advised the project team in dealing with emergencies in field work. The field survey would not be possible without the support of provincial and district governments in Madang and Morobe provinces. We appreciate the strong support and collaboration of the local governments in making the field work possible. The project benefited from the expertise and knowledge of the National Statistics Office (NSO) in survey implementation and supervision in PNG. The project team appreciate the support provided by the NSO team led by Mr. Roko Koloma, National Statistician and Late Mr. Peter Siopun, Assistant National Statistician, and joined by Mr. Tom Rabu, Statistical Officer, Ms. Dorothy Sapalojang, Manager of Business Statistics Branch, Ms. Francesca Tinabar, Manager of Policy and Research, Policy and Coordination Division, Ms. Annette Takaku—Statistician, Mr. Lohia Vaieke—Urban Operations Branch Manager, Mr. Vagi Guba, Statistical Officer, and Mr. Kevin Nelson, Web Master. With the meticulous efforts done by linguists and academia to standardize the translation of financial terms in the survey questionnaire into Tok Pisin for the first time, the survey was carried out in Tok Pisin with great success. The credit goes to Thomas Willie, Tom Rabu, Sandra Fore and Dean Woruba for translation and the Technical Review Committee for testing and finalization of the survey instrument in Tok Pisin: Patricia Passingan, Dorothy Sapalojang, Anna Irumai, Thomas Willie, Tom Rabu, Sandra Fore, Dean Woruba, and Rufina Peter. Special thanks go to Mr. John Mangos, Managing Director of Digicel PNG, and the Digicel staff for providing the wireless connection to allow data uploading from the field, which enhanced the data quality and reduced data processing time. We are especially grateful to the financial support and technical know-how provided by the Russia Financial Literacy and Education Trust Fund for developing the survey instrument and pilot testing, and by Korean Government Poverty Reduction and Socio- economic Development Trust Fund II for supporting the field survey implementation. In particular, we appreciate the strategic guidance provided by Richard Hinz, Program Manager of the Russian Trust Fund, and the insights and advice from Elaine Kempson on research methodology, in the design and development of the survey instrument. Finally, we would like to express our sincere gratitude to the individuals and households in Madang and Morobe provinces who participated in the survey and shared the information used for this report. vi Contents Tables, Figures, and Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x Abbreviations and Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1. Background 1.1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2. Conceptualising Financial Inclusion and Financial Capability . . . . . . . . 1 Financial Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Financial Capability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.3. Overview of the Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Development of the Financial Capability Survey . . . . . . . . . . . . . . . . . . 5 PNG National Financial Capability Survey . . . . . . . . . . . . . . . . . . . . . . . 5 Survey Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Survey Instrument and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Field Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 2. . . . . . . . . . . . . . . . . . . . . . 10 Financial Inclusion and Financial Services in PNG  2.1. Financial Inclusion in PNG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2. PNG Government Commitment to Increasing Financial Inclusion . . . . 10 2.3. Bank of Papua New Guinea Commitment to Increasing Financial Inclusion and Financial Literacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.4. Retail Financial Services in PNG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.5. Accessing Retail Financial Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Overview of Morobe and Madang 3.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 vii Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea 3.2. Population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Population Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Population Change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 3.3. Economy and Livelihoods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.4. Infrastructure and Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Telecommunications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Health . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 3.5. Financial Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4. Financial Inclusion in Morobe and Madang . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1. Facilitators of Financial Inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Confidence with Communicating in English . . . . . . . . . . . . . . . . . . . . . 34 Use of Mobile Phones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.2. Use of Formal Financial Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Number of Financial Products Owned . . . . . . . . . . . . . . . . . . . . . . . . . 36 Overview of Financial Products Owned . . . . . . . . . . . . . . . . . . . . . . . . 37 Payments and Remittances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Long Term Savings and Investments . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Credit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.3. Barriers to Financial Services Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Access to Financial Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Affordability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4. Use of Informal Financial Services . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Credit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 4.5. Responsibility for Selection of Financial Products . . . . . . . . . . . . . . . . 52 viii CONTENTS 5. Financial Capability in Morobe and Madang . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.1. Managing Current Household Cash Flows . . . . . . . . . . . . . . . . . . . . . . 54 Planning and Budgeting Household Cash Flows . . . . . . . . . . . . . . . . . . 54 Management of Household Expenditure . . . . . . . . . . . . . . . . . . . . . . . . 56 5.2. Planning Future Household Cash Flows . . . . . . . . . . . . . . . . . . . . . . . . 59 Planning for Major Future Expenditure . . . . . . . . . . . . . . . . . . . . . . . . . 59 Planning for Children’s Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Planning for Older Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 5.3. Responsibility for Household Financial Management . . . . . . . . . . . . . . 65 5.4. Financial Knowledge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Selection of Financial Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Understanding the Cost of Financial Services . . . . . . . . . . . . . . . . . . . . 68 5.5. Managing Relationships with Financial Institutions . . . . . . . . . . . . . . . 69 5.6. Findings Relative to Other Financial Capability Studies . . . . . . . . . . . . 69 6. Implications for Policy and Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Appendices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Appendix 1: Overview of Districts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Morobe Province . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Madang Province . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Appendix 2: Overview of Households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Respondents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Household Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Dwelling Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Appendix 3: Sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Appendix 3a: Sampling Notes for PNG National Financial Capability Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Appendix 3b: Sampling Weights for the PNG Financial Capability Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Appendix 4: Field Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Appendix 5: Glossary of Financial Terms in Tok Pisin . . . . . . . . . . . . . . . . . . . 121 Appendix 6: Literature Review of Financial Capability Measurement . . . . . . 125 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 ix Tables, Figures, and Maps Tables Table 1:  Commercial Bank Retail Banking Services . . . . . . . . . . . . . . . . . . . . . . . . . 13 Table 2:  Access to Financial Services Morobe and Madang . . . . . . . . . . . . . . . . . . 32 Table 3:  Ownership and Use of Mobile Phones . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Table 4:  Number of Financial Products Owned . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Table 5:  Financial Product Ownership by Category . . . . . . . . . . . . . . . . . . . . . . . . . 38 Table 6:  Households Incurring Expenditure by Type . . . . . . . . . . . . . . . . . . . . . . . . 39 Table 7:  Payment Modality for Expenses Incurred by the Household . . . . . . . . . . . 39 Table 8:  Households Receiving Receipts by Category . . . . . . . . . . . . . . . . . . . . . . . 40 Table 9:  Receipt Type for Income Received by the Household . . . . . . . . . . . . . . . . 41 Table 10:  Savings Account Ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Table 11:  Long Term Savings Ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Table 12:  Formal Credit Obligations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Table 13:  Estimate of Bank Account Costs as Percent Income for a Household on the Poverty Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Table 14:  Informal Sector Savings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Table 15:  Informal Sector Borrowing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Table 16:  Responsibility for Financial Product Selection . . . . . . . . . . . . . . . . . . . . . 53 Table 17:  Reasons for Household Cash Shortage . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Table 18:  Household Borrowing Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Table 19:  Planning for Children’s Future . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 Table 20:  Planning for the Children’s Future by Principal Source of Income . . . . . 61 Table 21:  Expected Means to Meet Expenses When No Longer Working Due to Age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Table 22:  Current and Expected Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 x Tables, Figures, and Maps Table 23:  Knowledge of the Cost of Financial Services . . . . . . . . . . . . . . . . . . . . . . 68 Table 24:  Principal Source of Household Income . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Table 25:  Principal Source of Individual Income . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Table 26:  Estimated Monthly Income by District Predicted Poverty Level . . . . . . . 85 Table 27:  Estimated Monthly Individual Income by Gender, Location and Livelihood Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Table 28:  Highest Level of Education in Household . . . . . . . . . . . . . . . . . . . . . . . . . 86 Table 29:  Land Ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Table 30:  Services and Access . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Table 31:  First Stage Selection of Provinces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Table 32:  Cross-Tabulation of Female Literacy Group and Poverty/Disadvantage Group Amongst Districts within the Selected Seven Provinces . . . . . . . . . . . . . . . . 97 Table 33:  Second Stage Selection of Districts in the Seven Provinces . . . . . . . . . . 98 Table 34:  Cross-Tabulation of Female Literacy Group and Poverty/Disadvantage Group Amongst Selected Districts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Table 35:  Selected Census Units (Outside of NCD) . . . . . . . . . . . . . . . . . . . . . . . . . 102 Table 36:  Substitute Census Units in Case of Inaccesibility of a Selected CU (One per District) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Table 37:  Selected Census Units in NCD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Table 38:  Districts in Morobe and Madang, by Stratification Groups (Selected Districts Shown in Italics) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Table 39:  Inputs into the Calculation of District-Level Weights for the Selected Districts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 Table 40:  Census Unit and Household Weights . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 Figures Figure 1:  Financial Capability Concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 Figure 2:  Microfinance Outreach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Figure 3:  Branch Density per 100,000 Adults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Figure 4:  ATM Density per 100,000 Adults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Figure 5:  Population Density on Arable Land (per sq. km) . . . . . . . . . . . . . . . . . . . . 19 Figure 6:  Average Household Size—2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure 7:  Age Profile—Years of Age (Percent) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 xi Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 8:  Age/Gender Breakdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 9:  Migrants 2011 (Percentage of Total Population) . . . . . . . . . . . . . . . . . . . . 22 Figure 10:  Interprovincial Migration 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 11:  Percentage of Employed Population in Subsistence Employment 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 12:  Percentage of Employed Population in Wage Employment—2011 . . . . 24 Figure 13:  Household Profile—Livelihood and Home Ownership—Percent 2011 . 25 Figure 14:  Nationwide Coverage by Digicel in 2011/2012 . . . . . . . . . . . . . . . . . . . . 28 Figure 15:  School Attendance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Figure 16:  Grade 10 Highest Grade (Percent of Citizens) . . . . . . . . . . . . . . . . . . . . 30 Figure 17:  Literacy 2011 (Percent of Citizens) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 18:  Confidence in Communicating in English . . . . . . . . . . . . . . . . . . . . . . . . 35 Figure 19:  Remittances Sent and Received . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Figure 20:  Bank Account Ownership and Distance to Nearest Branch . . . . . . . . . . 43 Figure 21:  Propensity to Save . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Figure 22:  Planning Horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Figure 23:  Knowledge of Spending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 Figure 24:  Expectation Strategies Will Cover All Expenses . . . . . . . . . . . . . . . . . . . 65 Figure 25:  Responsibility for Management of Household Expenditure . . . . . . . . . . 66 Figure 26:  Responsibility for the Management of Household Finances by Gender . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Figure 27:  Selection of Financial Products . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Figure 28:  Respondent Age Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 MAPS Map 1:  Papua New Guinea (Highlighting Morobe and Madang Provinces) . . . . . . . 18 Map 2:  Morobe and Madang Population Density . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Map 3:  Access to Service Centre . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Map 4:  Morobe Districts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 Map 5:  Madang Districts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 xii Abbreviations and Acronyms ADB Asian Development Bank AFI Alliance for Financial Inclusion BPNG Bank of Papua New Guinea CEFI Centre for Excellence in Financial Inclusion CU Census Unit HIES Household Income and Expenditure Survey LFI Licensed Financial Institution MEP Microfinance Expansion Project MFI Microfinance Institution NSO Papua New Guinea National Statistics Office PFA Household principal financial actor PFIP Pacific Financial Inclusion Programme PGK Kina PNG Papua New Guinea S&L Savings and Loans Society SME Small and Medium Enterprise SMK Salim Moni Kwik (remittance) Sunday-Sunday A form of informal ROSCA. Typically a small group which meets after the (fortnightly) payday to pool savings and make loans to group members WB The World Bank xiii Glossary Financial capability The internal capacity to act in one’s best financial interest, given socio-economic environmental conditions. It encompasses the knowledge (literacy), attitudes, skills and behaviours of consumers with regard to managing their resources, and understanding, selecting, and making use of 2 financial services that fit their needs. Financial inclusion A state in which all working age adults, including those currently excluded by the financial system, have effective access to the following financial services provided by formal institutions: credit, savings, payments, and insurance (GPFI & CGAP, ND). Financial literacy Knowledge of the financial products, services, practices and concepts required to make effective money management decisions. Formal financial sector Financial services provided by financial institutions supervised by the Bank of Papua New Guinea. Informal financial sector Financial services provided by financial institutions and individuals who are not supervised by the Bank of Papua New Guinea. For example informal microfinance and informal money lenders. Microfinance Financial institutions that target poor and low-income households. Financial Institution Any public or private institution whose main function is the provision of financial services for its customers or members. Money Lending An informal lender providing credit, usually short term, usually to individuals. Can also include loans by friends or relatives which need to be repaid. Principal Financial Actor A person who is responsible for making financial decisions on behalf of their household. 2  http://responsiblefinance.worldbank.org/~/media/GIAWB/FL/Documents/Publications/ Why-financial-capability-is-important.pdf. xiv Executive Summary Objective of the Report The study of financial inclusion and financial capability in Morobe and Madang provinces is the first population level study of financial inclusion and financial capability in Papua New Guinea. This report is an initial report of the national study of financial inclusion and financial capability in PNG. The objective of this report is to examine levels of financial inclusion of adults who make financial decisions on behalf of their households across the key financial product groups of savings, long term savings, and credit and protection products. The report also examines the financial capability of adults who make financial decisions on behalf of their households, particularly in respect to those aspects of financial behaviour which impact the management of household cash flows. The report has examined financial inclusion and financial capability for both women and men who make financial decisions on behalf of their household. Knowledge of levels of financial inclusion and related understanding of financial capability is currently very low in Papua New Guinea. Increasing our knowledge and understanding is particularly important as the formal financial system continues to develop rapidly in PNG and households, whether urban or rural, are required to use a broad array of financial instruments, manage complex household finances within an extended planning horizon which, increasingly, encompasses the need to provide for retirement. The report makes a significant contribution to increasing understanding and, whilst limited in geographic scope to Morobe and Madang provinces, provides an important input to the further development of financial inclusion and financial literacy strategy in Papua New Guinea. The study has been undertaken by the Bank of Papua New Guinea; field work was undertaken by the Papua New Guinea Institute for National Affairs, with the support of the National Statistics Office. Technical assistance was provided by the World Bank. The study was funded by the Korean Poverty Reduction and Socio-Economic Development Trust Fund II. xv Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Structure of the Report The report is structured in six chapters. Chapter 1 discusses the background to the study. The concepts of financial inclusion and financial capability are briefly discussed. Following this, the development of the survey (within the context of the development of the global World Bank Financial Capability survey instrument) is overviewed. The study objectives, instrumentation, methodology, sampling and field-work are also briefly discussed. Chapter 2 discusses financial inclusion and financial capability in PNG generally. Increasing levels of financial inclusion and enhancing the financial capability of the population are key strategic objectives of the Bank of Papua New Guinea and policy objectives for the government of Papua New Guinea. The retail financial services landscape in PNG is briefly overviewed. Chapter 3 provides an overview of Morobe and Madang. Financial inclusion and financial behaviour do not stand alone but are a function of the economic environment, the infrastructure and public services and the education environment. Morobe and Madang provinces in some ways entail a microcosm of many of the diverse geographical and social characteristics of Papua New Guinea. Whilst it is impossible for any province to represent the entire country’s range of physical and human characteristics, Morobe and Madang do reflect some of that diversity and the related challenges facing the extension of formal financial services in PNG. The current financial services environment in Morobe and Madang is overviewed. Chapter 4 examines financial inclusion in Morobe and Madang. The chapter commences with an examination of two facilitators of financial inclusion: confidence in communicating in English and access to and use of a mobile phone. The use of formal financial services is examined across the key product groups of payment services, savings, long term savings (including provident and superannuation), credit and protection (insurance) products. Barriers to the use of financial services are examined, in particular access barriers, gender and affordability. The use of informal financial services is examined. Chapter 5 examines financial capability in Morobe and Madang. Five aspects of financial capability are explored. The chapter commences with an examination of the management of current household cash-flows, from the perspective of both planning and budgeting cash-flows and the management of flows. Planning future household cash-flows is then examined. Three aspects of future cash-flow management are explored: the planning for major future expenditure, planning for the future of children in the household and planning for older age. Responsibility for household financial xvi Executive Summary management is examined. Following this the chapter examines relationships with financial services providers and the process used to select financial products and services and understanding of the cost of financial services. The chapter concludes with a brief discussion of the findings from the present study relative to other financial capability studies, in particular studies in PNG and other Pacific Island countries. Chapter 6 discusses the policy and strategy implications of the study findings, with a particular focus of the implications for the achievement of the Maya Declaration goals. The Annexes provide detailed information on the survey instruments, the sampling methodology, an English-Tok Pisin Glossary of Financial Terms, and literature review of financial capability measurement. Key Findings Financial Services Access There is good access to formal financial services access points in urban and township locations. Retail banking services, including ATM and EFTPOS services are available at multiple locations in Lae and Madang and also in townships (for example Bulolo and Finschafen). Savings and Loan and microfinance institutions also have outlets, primarily in urban locations. In addition to accessibility to financial services, households in urban and township communities can access a broad range of formal financial services. A range of payment services, savings, long term savings and investment services and credit services (including asset finance) are offered by multiple financial services providers. Consumers in urban locations appear to have both access and choice. The situation in rural communities is very different. Only one bank, BSP, has a rural agent network. The number of rural agents, relative to the rural population is, however small. A very significant proportion of the rural population in Morobe and Madang effectively has no, or very limited access to formal financial services. In addition, the range of financial services available to rural communities is very limited. Products and services are limited to those which can be offered by agents. Overall, it appears many consumers in rural locations in Morobe and Madang have neither access to formal financial services or a set of financial products and services from which to select an appropriate product. Financial Inclusion A very high percentage of rural respondents (60–80 percent) owned no financial products. Women were more likely to report owning no financial products than xvii Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea men. By contrast in urban communities men reported owning on average 3.4 financial products and women 1.3. Respondents whose principal source of income was formal sector wages or salary were more likely to report financial product ownership. Payments for day-to-day purchases were made using cash, irrespective of whether or not the respondent had a bank account with debit card access. Items which were commonly paid for by electronic channels were school fees, formal loan repayments and rent/lease payments (in urban households). Most wages or salaries were typically received by credit to a bank account. By contrast payment for sales from informal activity or self-employment was almost exclusively cash based. In urban locations levels of account ownership appear to be similar to those reported by other developing countries. In rural locations, by contrast, 21 percent of men and 9 percent of women owned a savings account. There is some evidence of multiple account ownership. Twenty one percent (21 percent) of respondents who reported owning a debit card based savings account also reported owning a passbook savings account. Given the differences in the product range between banks and S&Ls/microfinance, this suggests approximately 20 percent of bank savings account customers may also have an account at another institution. Respondents who reported owning a savings account also reported money was kept safe in the account. By contrast, respondents who did not have an account reported keeping money safe by hiding it or using a locked box. Households that owned a savings account were more likely to report the household tried to save money for the future, tried to save regularly and tried to keep funds aside to provide for emergencies or unexpected expenses. Use of informal savings appears to be related to employment. Most respondents who reported having informal savings associated the product with employment. Urban respondents in formal sector employment were more likely to report long term savings, in particular provident/superannuation fund membership. Approximately 7 percent of adults reported some form of formal borrowing. This is only slightly lower than levels found in other developing countries. Men living in urban communities, whose principal source of income was formal sector wages and salaries were the most likely group to report having formal credit commitments. Levels of insurance product ownership were very low. Even in urban households only 8 percent of households reported having insurance. Rural households were significantly more likely to report informal borrowing, relative to use of formal credit, than urban households, although overall more urban households reported the use of informal credit than rural households. Reported levels of borrowing from money lenders (6 percent) were only slightly higher than levels reported in other developing countries. xviii Executive Summary Barriers to Financial Inclusion The principal barrier to financial inclusion appears to be the limited ability of most adults in Morobe and Madang to access financial services. Generally, as noted above, rural communities do not have access to formal financial services. Overall, the further the household is from an urban area the more likely the household is to be financially excluded. Women appear to be significantly more likely to be financially excluded than men, particularly women living in urban communities. Globally 58 percent of women report owning an account, compared to 65 percent of men, a seven percentage point difference. As discussed, in rural communities levels of account ownership are very low for both men and women. In urban communities 38 percent of women and 68 percent of men reported owning an account, a thirty percentage point difference. Most urban respondents, in particular men, were confident of their ability to communicate in English, both spoken and written communication. By contrast most rural respondents, in particular women, stated they could not communicate in English. This may be a barrier to the informed use of financial services, given most documents are written in English. There is a very significant difference in mobile phone ownership or access, and usage, between urban and rural communities. Rural women in particular appear to be at a significant disadvantage in respect to the opportunity to use a mobile phone for financial services. Without further development of mobile phone access and the capability to text, the opportunity to use the mobile phone as a primary channel for retail savings and transaction services may be limited, at least in some rural areas. Financial Capability Cash Flow Management Most households do not plan or budget but prioritise major expenditure items. Whilst most households reported planning how income would be used, households also reported most plans and budgets were general, covering only major items, and were not written down or usually adhered to. Slightly over half of households reported having left-over money after meeting household expenses at least some of the time. Households reported saving cash surpluses to cover unforeseen expenses or to pay for food or other necessary items. By contrast 80–90 percent of households reported running short of money after meeting household expenses. Cash flow shortfalls were funded by borrowing from family or friends (in urban households money was also borrowed from informal money lenders or an employer), selling something (in rural households) or simply going without. xix Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Longer Term Financial Planning Many households appear to plan their income cycle. A significant number of households also reported planning at least six months ahead. Seventy percent (70 percent) of households stated they would not be able to cover a major unexpected expense (equivalent to approximately one month’s income) without borrowing. Most households reported planning for their children’s future. Households typically used more than one approach to planning for the future, the common being to save money, in particular for the children’s education. Forty four (44) percent of respondents reported they had no strategy to provide for when they were no longer able to work owing to old age. Most respondents who did have a strategy reported they expected to earn income from a business. Respondents who were employed in the formal sector also cited long term savings, in particular provident or superannuation. Respondents who earned formal sector income were more likely to consider they would be able to meet expenses when they were no longer working. Less than 50 percent of those earning informal sector income or self-employed considered they would be able to meet expenses. Approximately one quarter of respondents stated they expected to receive support from family or kinship groups. Responsibility for Household Financial Management Men were significantly more likely to cite they are responsible for the management of household expenses than women. Overall, the further out the expenditure horizon, or the less certain the expenditure (for example requests for financial assistance), the more likely that respondents reported no one in the household was responsible for the management of that category of household expenditure. Financial Knowledge Urban households appear to be generally more aware and focussed on the household’s financial situation than rural households. Overall, however, most respondents stated they considered they were disciplined at managing money and only rarely or occasionally bought unnecessary items before buying necessary items or when they knew they could not afford the item. Selecting Financial Products and Managing Relationships with Financial Services Institutions Respondents who were confident in their ability to communicate in English were more likely to search for information from a range of sources and to consider alternatives before purchasing a financial product. Respondents who could not communicate in xx Executive Summary English were significantly less likely to check terms and conditions before committing to purchase a financial product. Knowledge of the cost of money was very limited. Most respondents (80 percent+) did not know how much interest had been paid on loans or received on deposits and did not know the fees which had been paid over the previous year. Policy Implications Overall Implications The findings from the study suggest that the current National Financial Inclusion and Financial Literacy Strategy may need to be further developed, in particular to facilitate significant strengthening of the rural financial services architecture. This is a pre-condition for enhancing rural financial inclusion. The findings also indicate significant strengthening of financial inclusion and financial literacy programmes, in particular urban programmes, will be required in order to increase financial inclusion by women. Levels of financial knowledge, in particular knowledge of the cost of financial services suggest a continuing commitment is required to further strengthen consumer protection. Implications for the Achievement of the Maya Declaration Goals The implications of the study findings for the achievement of the Maya Declaration Goals are discussed within the context of each Goal: Goal: To reach one million more unbanked low-income people in Papua New Guinea, 50 percent of whom will be women Situation in Morobe and Madang Levels of financial inclusion in urban communities in Morobe and Madang provinces may be approaching, or may be at levels found in other developing countries. Urban households, in particular households in which the principal source of income is formal sector wages or salary, appear to be engaging with the formal financial system across a broad range of product groups, extending beyond savings/transaction accounts to long term savings and formal credit. Levels of financial inclusion in rural communities, across all product groups are, however, very low. Women in urban communities are significantly more likely to have a bank account than women in rural communities and may be accessing financial services indirectly by accessing a bank account owned or controlled by a male. Relative levels of financial exclusion by women living in urban communities are significantly higher than those found in rural communities. xxi Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Implications for Policy and Practice Urban communities have access to a range of financial service providers and financial services. However, many rural communities have, effectively, very limited or no access to formal financial services. Findings from Morobe and Madang suggest that, without major development of the rural financial services infrastructure, bringing large numbers of Papua New Guineans, who are currently financially excluded, into the formal financial system will not be possible. Unless there is a substantive change to financial inclusion products/services/ programmes to bring women into the formal financial system, the goal of gender equity will probably not be achieved and the gender gap may be further exacerbated. Goal: (BPNG) to lead efforts to create a financially competent generation of Papua New Guineans through financial education and financial literacy Situation in Morobe and Madang Most households in Morobe and Madang try to plan and budget cash-flows. However, for most households’ plans and budgets are informal, cover major items and are not documented. Effectively most households only prioritise spending. Overall, urban households, in particular households which receive regular wages or salary, are more likely to pro-actively manage household finances than rural households. Women are less involved in all aspects of household financial management than men and have less responsibility for the management of household finances than men. Many adults, in particular rural households and women, do not know how they will fund their own and their household’s expenses when they are no longer working. Implications for Policy and Practice The focus on budgeting and planning, which is a common feature of financial literacy programmes, would appear to be appropriate. As the use of formal credit expands, inclusion of responsible borrowing modules in financial literacy programmes may also be warranted. Consideration may need to be given to including longer term savings, or asset accumulation modules, in order to increase awareness of the need to prepare for older age and how gradual longer term savings can be achieved. There is an ongoing need for financial literacy programmes specifically for women. Goal: To actively support innovative use of technology for scaling-up access to financial services and financial literacy xxii Executive Summary Situation in Morobe and Madang Branch, ATM and EFTPOS networks in Morobe and Madang provinces are largely urban. Agent networks do not, at this time, appear to be sufficiently dense to enable most rural households to readily access formal financial services. However, in rural communities, levels of mobile capability (levels of phone ownership combined with adults’ ability to use the mobile phone) are low. Implications for Policy and Practice Other than BSP Rural, commercial financial services providers are not expanding services into rural areas but are, not unexpectedly, focusing on urban communities. It may be appropriate to consider the development of financial services delivery architecture for rural communities in PNG. Developing extensive branch networks in rural communities is unlikely to be economic. Technology based solutions are likely to be a key component of a rural financial services architecture. However, the combination of more limited ownership or access to a mobile phone and mobile phone capability in rural communities, in particular by women, suggests that the expansion of mobile phone-based financial services (as opposed to the use of mobile telephony to deliver agent based financial services) in rural areas may require concurrent capacity building. Goal: To strengthen consumer protection by issuing prudential guidelines and creating a platform for various national regulators and industry networks to monitor consumer protection Situation in Morobe and Madang Financial services consumers appear to have a limited understanding of the cost of the financial services they use. Many consumers, in particular consumers who are not confident in communicating in English, do not search for information about financial services before committing to buy a product, they do not consider alternatives or look at alternative products and, perhaps most importantly, they frequently do not check terms and conditions of the products they purchase. Implications for Policy and Practice Consumer awareness appears to be limited for many adults and there is therefore potential vulnerability to predatory practices. Few households reported a dispute with a financial services institution. Nevertheless a platform to monitor consumer protection may be warranted. xxiii Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Goal: To promote regular collection and use of financial access data to inform policy making and help identify key dimensions of financial inclusion in Papua New Guinea Findings from the present study have contributed to the understanding of financial inclusion and financial capability in PNG. However, the findings are not national. It is recommended funding be sought to complete the survey to obtain a national view on Financial inclusion and Capability in Papua New Guinea. Limitations of the Report and Recommendations This report is an initial output from the national financial capability study, and focuses on Morobe and Madang provinces only due to both financial and logistic constraints to conduct a national survey at this stage. The focus of the study is households. The use of financial services by formal and informal businesses has not been examined. The relationship between financial inclusion and financial capability has been examined by financial inclusion and financial capability studies in several countries using the World Bank Financial Capability Survey instrument. As the present study is an initial output from the national study, it has not sought to examine this relationship fully. It is appropriate to examine the relationship between financial inclusion and financial capability once the national survey has been completed. It is intended that the study will, in time, progress to become a nationally representative study of financial inclusion and financial capability in PNG. The sampling for the present study is representative at the combined population level for Morobe and Madang. Findings cannot be disaggregated to Morobe and Madang separately. In addition, whilst potentially indicative of likely findings from the national study, the findings from the present study cannot be interpreted as nationally representative. The report is point in time. The economic and financial landscape in PNG is evolving rapidly and regular follow up studies will be required to maintain the currency of the data. It is thus recommended that the national survey be carried out based on the learning from the regional survey done in Morobe and Madang provinces within the next two years to establish the national mapping of financial capability of PNG population. The survey instrument has been developed and tested in Tok Pisin and can be further adapted and simplified based on the feedbacks from this regional survey. The sampling methodology was established for the national survey and can be readily used for carrying out the national survey. The field implementation management was fully developed and tested and capacity of the field survey teams has been enhanced through training and on-site learning. The data collection methodology has also been piloted and tested and can be replicated for future national surveys. xxiv 1.  Background 1.1. Introduction This report contributes to the developing understanding of financial inclusion and financial capability in PNG. The PNG National Financial Capability study is the first national study of financial inclusion and financial capability in PNG and is one of a small number of studies undertaken in PNG seeking to develop an understanding of financial inclusion, financial knowledge and skill and financial behaviour. The present report is an initial output from the national study, focusing on Morobe and Madang provinces. It is intended that the study will, in time, progress to become a nationally representative study of financial inclusion and financial capability in PNG. 1.2. Conceptualising Financial Inclusion and Financial Capability Financial Inclusion 3 Over recent decades, financial inclusion has emerged as a global priority. Financial inclusion is an enabler of effective, sustainable, participation in the contemporary money economy. Financial inclusion is an issue in PNG. It is estimated that currently in PNG up to 85 percent of the population does not have access to a bank account. However, financial inclusion, whether in a developed or developing country context, extends beyond basic deposit services. In a developed country context, for example, the UK Treasury has defined financial inclusion as: “Access to appropriate financial services so that people can manage their money effectively, securely and confidently on a day-to-day basis; plan for the future and cope with financial distress to protect against short term variations in income and expenditure and take advantage of longer term opportunities and deal effectively with financial distress.” (HM Treasury, 2007, p. 25) 3   Refer for example: http://www.gpfi.org/. 1 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea In a developing country context CGAP has defined financial inclusion as: “A state in which all working age adults, including those currently excluded by the financial system, have effective access to the following financial services provided by formal institutions: credit, savings, payments, and insurance.” (GPFI & CGAP, ND) This report examines access to, and use of, a broad range of financial products and services by households in Morobe and Madang provinces; with a particular focus on understanding location based differences and gender differences in the use of financial services. Levels of financial inclusion found in this study can be compared, at a product level, to levels of financial inclusion found in other Pacific island countries and at a product- category level, to levels of financial inclusion in other countries generally. There is considerable homogeneity of product attributes across regions. Whilst financial services fees and value-added services differ widely between countries, core product functionality is relatively consistent globally. For example, savings products, card- based transaction products, credit cards, consumer loans and term investments have broadly similar features in most countries. There is significant homogeneity of product functionality in financial products available across Pacific island countries. This is not surprising perhaps given the same group of commercial banks operate in most Pacific island countries. Financial Capability Financial capability has been defined by the World Bank as: “The (internal) capacity a person has to act in their best financial 4 interest, given socio-economic environmental conditions.” Financial capability encompasses the knowledge, skills, behaviours and attitudes people employ to manage their resources; their understanding of financial services; and the selection and use of financial products and services to meet their needs. Financial capability is a broader concept than financial literacy which focuses on knowledge of the financial products, financial services, financial practices and financial concepts. Financial capability can be conceptualised as spanning four domains: Base skills, financial knowledge, financial attitudes and financial behaviours. The key components of each domain are shown in Figure 1. 4  http://responsiblefinance.worldbank.org/~/media/GIAWB/FL/Documents/Publications/ Why-financial-capability-is-important.pdf. 2 Background Figure 1  Financial Capability Concepts Base Skills Financial Knowledge Financial Attitudes Financial Behaviours ■■ Numeracy skills ■■ Knowledge of ■■ Reasons for or ■■ Money financial concepts for not saving, management ■■ Literacy skills (inflation, borrowing, (managing day-to- compound interest investing, etc. day finances) etc.) ■■ Attitudes towards ■■ Long-term planning ■■ Awareness of the future (preparing for financial products emergencies and ■■ Confidence in own and services retirement) plans for old age ■■ Practical know- ■■ Financial decision- ■■ Attitude toward how (how to make making (ability to budgeting, saving, payments, how choose appropriate lending (etc.) to open a bank financial products) account etc.) ■■ Seeking financial advice This report examines the financial capability of adults in Morobe and Madang who make financial decisions on behalf of their households. Financial capability is a situated construct. For example, the financial capability required by a subsistence farmer living in a remote rural community is likely to be different than the financial capability required by an employee of a large corporation living in a metropolitan environment. Required levels of literacy and numeracy are likely to differ. The level of financial knowledge required by a person who receives a regular salary and who has a range of borrowing will be different than that required by a subsistence farmer whose cash income is intermittent and who has no formal borrowing. Lifestyle differences, individual attributes and differences in the use of financial products and related levels of required financial knowledge and skill will impact financial attitude. The core elements of financial behaviour are, however, to some extent constant: households engaged in the money economy need to be able to manage current household cash-flows, plan for the future, make decisions about financial products (whether formal or informal) and seek advice about money and finance. Studies of financial capability in several countries have included the construction of a financial capability scale using factor analysis. Whilst the factors and loadings for each individual country scale are unique, the development of a scale enables comparison between relative levels of financial capability and financial inclusion within jurisdictions and the comparison of levels of required financial capability across jurisdictions. 3 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea The present study has not constructed an index of financial capability. There are three reasons for this: Firstly, this report is an interim report examining financial inclusion and financial capability in two provinces of the Momase Region. The national study has yet to be completed and it is therefore not possible to construct a national index. Secondly, aspects of financial capability will require further testing before being used to construct an index, in particular attitudinal variables. PNG has three official languages: Tok Pisin, Hiri Motu and English. English is the language of government and business. Hiri Motu is spoken in Papuan regions but not nationally. Tok Pisin has become widely used across PNG as a bridge across the great diversity of languages and dialects. Tok Pisin began as a trading language (‘broken English’) and continues to develop as a discrete language. However, there is, at present, no formal grammar for Tok Pisin and translating complex constructs effectively into Tok Pisin can be difficult, in particular translation of abstract concepts. Nevertheless, the study has translated attitudinal questions from the core World Bank Financial Capability instrument and responses are reported (with caution). Further testing will be required to determine the validity of including psychological/attitudinal questions in the financial capability scale. Thirdly, evidence from this study suggests the components of financial capability may differ very significantly between urban and rural communities and, therefore, it may not at this time be appropriate to develop a single national index of financial capability for PNG. Differences in levels of financial inclusion and the use of money between urban and rural communities revealed by the present study suggest it may be more appropriate, at least initially, to develop separate urban and rural indices. Several studies of financial inclusion and financial capability using the instrument developed by the World Bank administered Russia Trust Fund for Financial Literacy and 5 Education (the instrument used for this study) have tested respondents’ numeracy and literacy skills. It was decided not to test numeracy for the PNG National Financial Capability study. Given high levels of illiteracy, a numeracy test would need to be administered verbally and could be difficult to translate. For similar reasons, the study has not tested respondents’ knowledge of common financial concepts (for example inflation and compound interest) due to language constraints. There is, for example, no construct in Tok Pisin for ‘compound interest’. Efforts to translate into Tok Pisin resulted in the concept being explained within the question, which rather defeats the purpose of the question. Other financial capability studies which have tested literacy, have only tested comprehension rather than all aspects of literacy. Tok Pisin is a 5  https://www.finlitedu.org/. 4 Background verbal language and testing written comprehension is not appropriate (at least at this time). Testing English comprehension would require a formal literacy test which is not appropriate as English is not used widely in most of the households participating in the study. Instead, given the widespread use of English in financial services documents, respondents were asked how confident they were in communicating in English (verbal and written) with a bank branch or a government department. As a consequence of the foregoing, the financial behaviour of adults who make financial decisions on behalf of their households and related levels of financial inclusion are the principal focus of this interim report. 1.3. Overview of the Survey BPNG has supported three studies to increase understanding of financial inclusion 6 and financial capability in PNG. The present study has its origins in the participation by PNG in the development of a global instrument to measure financial capability. Development of the Financial Capability Survey In 2010 teams from eight countries (Papua New Guinea, Zambia, Malawi, Namibia, Tanzania, Uruguay, Mexico and Colombia) met to begin the development of a global instrument to measure financial capability. The development of the instrument was undertaken under the auspices of the World Bank, with funding provided by the 7 Russia Trust Fund for Financial Literacy and Education. The scope of the instrument and the instrument development methodology were agreed at the initial meeting. Development of the instrument was completed in 2011. The instrument was then extensively tested in the context of low and middle income countries. The Financial Capability survey is currently used or planned to be used in 14 countries in Latin America, Africa, Middle East and East Asia and the Pacific. PNG National Financial Capability Survey PNG piloted the financial capability survey in 2012. It was determined the survey could be successfully deployed in PNG, using a Tok Pisin translation, in both high income and low income households. Preparation for the PNG National Financial Capability Survey commenced in 2013 with funding provided by the Korean Poverty Reduction and Socio-Economic Development Trust Fund. 6  The other two studies are: 2013 Financial Diaries Study in Port Moresby, Goroka and Kimbe; 2013 Financial Competence study in Port Moresby, Mekeo and Galley Reach districts. 7  https://www.finlitedu.org/. 5 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea A national sample was developed with census units selected from five provinces. Field work for the survey commenced in 2014. Due to logistic and funding constraints it was determined the survey would be deployed initially only in the Momase Region. Two provinces had been selected for field work in Momase: Morobe and Madang. Morobe is the largest and one of the most diverse provinces in PNG. Lae, the Morobe provincial capital, is an accessible central location to train enumerators. Madang province is proximate to Morobe. Momase was considered to be an appropriate region for initial survey deployment and would enable the refinement of the fieldwork methodology for subsequent rounds of data collection. The financial capability survey instrument is a global instrument and data from each country is comparable. Minor adaptation is required for individual countries (primarily in respect to financial products). Two versions of the survey were developed by the World Bank Project Team: a household version and an individual version. As the objective of the PNG study was to understand household financial inclusion and the financial capability of adults who make financial decisions on behalf of their household, the individual version of the survey was not used. Enrolment for the generic financial capability survey is based on interviewing a single respondent in each household. This approach was not used for the PNG survey as a principal focus of the surveys supported by BPNG has been to develop a gendered understanding of differences in the level of financial inclusion and financial capability. Both the woman and the man who make most financial decisions on behalf of their household were interviewed. Separate interviews were conducted. Survey Objectives The objective of the National Financial Capability survey is to develop a baseline measure of financial inclusion and financial capability for PNG. This will enable, for the first time, an accurate assessment to be made of the situation in respect to financial exclusion in PNG—not simply in respect to bank account ownership, but across the range of products: transaction, savings, long term savings and credit, commonly used by households to manage and develop household resources. This will enable the development of a financial services architecture which can facilitate reducing levels of financial exclusion. The baseline study will also enable the relationship between financial capability (or perhaps more accurately financial capabilities) and financial inclusion to be better understood. This will facilitate the development and targeting of financial literacy programmes and consumer education. The objective of the financial capability survey of Morobe and Madang has been firstly, to commence data collection for the national survey, and secondly, to provide an initial insight into levels of financial inclusion and patterns of product ownership 6 Background in urban and rural communities and by women and men, and to develop an initial understanding of financial capability, in particular financial behaviour, to provide input to the further development of financial inclusion and financial literacy strategy in PNG. Survey Instrument and Methodology The PNG National Financial Capability Survey has used the generic financial capability instrument developed by the World Bank Project Team. Several questions have been added to meet local requirements, for example questions to determine respondents’ confidence with the use of English and mobile phone usage. The survey comprises three elements: Location Fact Questionnaire The location fact questionnaire was administered during enrolment by interview with a local community leader. The questionnaire collects information in respect to the education, health and financial services available at the location, or the distance required to travel to services (and the mode of transport commonly used to travel), along with the source and quality of electricity and water. Enrolment Questionnaire The enrolment questionnaire was completed during the enrolment of the household and comprised a household roster, information in respect to the principal dwelling used by the household and household durables available for use by members of the household. Financial Capability Survey The financial capability survey collected data covering a broad spectrum of elements of financial capability and financial inclusion: Management of remittances and payments, planning and management of current cash-flows and future cash-flows (both expected and unexpected), planning for old age, planning for the future of children living in the household, financial products and services used by the household and how these were selected, and understanding of the cost of financial services and relationships with financial services providers. A standard closed question survey was used. Questions were asked in either Tok Pisin or English (depending on the respondent’s preference). All interviewers were bilingual. Forward and back translation was undertaken, with an independent post-translation panel review by a reference panel of bilingual subject matter experts. Both the English 7 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea and Tok Pisin versions of the survey were tested in the field. The translation was also work-shopped with the enumerators. A glossary of financial terms was developed and translated to Tok Pisin (refer to Appendix 5 for the Glossary). An added complication was the need to ensure the questions could be understood by respondents who may have limited formal education. A test of the reading age required to comprehend the survey at first reading in English (Flesch-Kincaid) indicated the survey required a comprehension level approximating that which would be achieved at the completion of primary/early secondary education. Whilst it is not possible to test the comprehension level in Tok Pisin, it was considered reasonable to assume that the level of comprehension would, in general terms, be able to be preserved in the translated version. Respondents were required to answer all (relevant) questions. Each question allowed for refusal. Interviews were conducted at a location and at a time suitable to the respondent. A female and male interviewer visited the household. Women were interviewed by women and men by men. Interviews were confidential. Sampling National Sample When fully implemented, the National Financial Capability survey will be representative of the adult population in PNG. Sample selection has been based on a stratified four-stage random sample design, with preliminary counts from the 2011 Census of Population providing the sampling frame. Stratifying according to female literacy and predicted poverty, the first two stages selected 19 Districts from seven provinces, with probability proportional to estimated size. The provinces include three from the Highlands (Western Highlands, Jiwaka, Eastern Highlands), two from Momase (Madang and Morobe), and one each from the New Guinea Islands (East New Britain) and Papuan (Central) regions. Additionally, the National Capital District (NCD) is a separate survey strata. Within each District, five Census Units (CUs) were selected with probability proportional to estimated size (PPeS). Within each CU, ten households are selected by the interview teams, using circular systematic sampling. This yields an overall sample of 50 households per District and 950 households in total. A further 150 households will be surveyed from the National Capital District (NCD), with a target of six households per CU, in 25 CUs that have been selected with a PPeS approach. Combining the strata, a total of 1100 households will be surveyed, and in each household an adult male and adult female will be interviewed. Since not all households will have both an adult male and an adult female, the final sample size will be less than 2200 but should be approximately the sample of 2000. 8 Background Morobe and Madang As discussed, due to funding and logistical constraints the scope of the field work was contracted to Morobe and Madang provinces. The census units and districts selected for Morobe and Madang were not resampled. The selected census units were surveyed and the completed surveys were re-weighted to reflect the completed surveys and the revised scope. Data was collected from 8 Districts and 36 census units (4 census units could not be accessed due to environmental conditions). Three hundred and fifty nine (359) households from the sampled census units participated in the survey and 688 individual surveys were completed. Please refer to Appendix 3 for the sampling note prepared by the sampling consultant Dr. John Gibson in respect to the national sample and the subsequent note discussing the re-weighting for Morobe and Madang. Please refer to Appendix 1 for a brief overview of the districts in Morobe and Madang in which field-work was undertaken. Please refer to Appendix 2 for a brief overview of the households interviewed in Morobe and Madang. Field Work Field work for the survey was undertaken by the PNG Institute for National Affairs (INA), with support from the National Statistics Offices in Port Moresby, Lae and Madang. Enumerators were recruited locally and were trained by the BPNG, INA and World Bank project team members and senior NSO staff. Field work was undertaken between July and December 2014. The Location Fact Questionnaire and the Enrolment Questionnaire were completed using paper based survey forms. Data entry was undertaken by INA. The Financial Capability survey was completed using off-line tablet based data collection. The tablets were provided by BPNG and mobile phone wi-fi data upload capacity was provided by Digicel. Please refer to Appendix 4 for a timeline of the field work and key activities. 9 2.  Financial Inclusion and Financial Services in PNG Improving access to financial services continues to be a very significant challenge for PNG. There are significant barriers to increasing financial access, in particular operational challenges and infrastructure weaknesses which result in high provider costs, and the limited financial capability of clients (IMF & World Bank, 2011). 2.1. Financial Inclusion in PNG Estimates of the number of adults who have accounts with regulated financial institutions vary between approximately 435,000 and 800,000 adults. ADB estimated approximately 15 percent of the adult population (approximately 600,000 adults) is 8 included in the formal financial sector (ADB, 2008). The 2011 Financial Services Sector Assessment (IMF & World Bank, 2011) estimated between 500,000 and 800,000 people were included in the formal financial system. BPNG (Anderson, Kunjil, Ngodup, & Tongia, 2013) estimated the number of account holders at regulated financial institutions to be 435,316, and that approximately 7 percent of the adult population had at least one loan with a regulated financial institution. BPNG estimates the unmet demand for deposit services from the economically active adult population to be approximately 5.38m people (Anderson, et al., 2013) and the unmet demand for credit to be between 54–64 percent of the economically active adult population. 2.2. PNG Government Commitment to Increasing Financial Inclusion Financial inclusion is a development priority for the Government of PNG. Expanding financial inclusion and enhancing financial literacy have been incorporated in key national plans: The Development Strategic Plan 2010–2030 (Department of National Planning and Monitoring, 2010) and Vision 2050 (National Strategic Plan Taskforce, 8   The present study found 15 percent of the adult population in Morobe and Madang reported owning a bank account. 10 Financial Inclusion and Financial Services in PNG 2011). The National Informal Economy Policy (2011–2015) identified financial inclusion as a priority action area to facilitate the development of the informal economy in Papua New Guinea and stated increasing levels of financial inclusion should be a major economic policy objective (DfCD & INA, 2011).  2.3. Bank of Papua New Guinea Commitment to Increasing Financial Inclusion and Financial Literacy The Bank of Papua New Guinea (BPNG) has taken the lead role in increasing financial inclusion and financial literacy in PNG. Financial inclusion is one of BPNG’s two strategic goals (BPNG, 2011). BPNG is a member of the Alliance for Financial Inclusion 9 (AFI) and signed the Maya Declaration in 2013, committing to seven financial inclusion and financial literacy goals (AFI, 2013). 1. To reach 1 million more unbanked low-income people in Papua New Guinea, 50 percent of whom will be women 2. To lead efforts to create a financially competent generation of Papua New Guineans through financial education and financial literacy 3. To actively support innovative use of technology for scaling-up access to financial services and financial literacy 4. To strengthen consumer protection by issuing prudential guidelines and creating a platform for various national regulators and industry networks to monitor consumer protection 5. To begin the process of integrating financial inclusion in local and national government, including getting the National Executive Council to endorse the National Financial Inclusion and Financial Literacy Strategy by quarter 4 of 2013 6. To promote regular collection and use of financial access data to inform policy making and help identify key dimensions of financial inclusion in Papua New Guinea 7. To optimize these results through knowledge sharing and effective coordination of stakeholders, including development partners, by the newly established Centre of Excellence for Financial Inclusion chaired by the Bank of Papua New Guinea. 9  http://www.afi-global.org/sites/default/files/publications/maya_declaration_bank_of_papua_ new_guinea.pdf. 11 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea The Maya Declaration goals provide a foundation for financial inclusion and financial literacy policy and strategy. In 2013, with the support of the Pacific Financial Inclusion Programme, BPNG brought together a broad cross-section of stakeholders to identify regulatory and market conditions necessary to encourage innovation and scale in 10 financial inclusion and financial literacy. The workshop provided a key input to the development of the National Financial Inclusion and Financial Literacy Strategy 2014–2015 (BPNG, 2013). The principal objective of the Strategy is to implement PNG’s seven commitments in the Maya Declaration on Financial Inclusion. BPNG is currently implementing several key strategic action items: 1. BPNG has sponsored and led the establishment of the Centre for Excellence in Financial Inclusion (CEFI). The primary objective of CEFI is to coordinate and monitor the implementation of the National Financial Inclusion and Financial Literacy Strategy. 2. BPNG has supported (in conjunction with ADB) the establishment and management of the Microfinance Expansion Project (MEP). The MEP is seeking to both strengthen the supply of microfinance services and increase the demand for microfinance services. 3. BPNG has committed to increasing financial inclusion data, in particular supply-side data. 2.4. Retail Financial Services in PNG Financial services accessibility has significantly increased over the last two years in Papua New Guinea as a result of financial inclusion programs rolled out by the Bank of PNG in collaboration with the partner financial institutions. Financial services access points has increased from 9,257 in December 2012 to 11,015 in March 2015, an increase of 13.51 percent due to expansion of Agents, ATMs, branches, sub-branches 11 and EFTPOS by financial institutions. There are a broad range of financial services institutions providing retail financial products and services in Papua New Guinea, including in Morobe and Madang. Savings and Loan Societies (S&Ls) are the earliest established retail financial institutions in PNG. S&Ls are mutual societies established by people ‘sharing a common bond of membership’ (for example common location or industry). S&Ls 10  http://www.afi-global.org/news/2013/8/28/bank-png-reveals-basis-its-upcoming-maya- declaration-commitment. 11   Source: BPNG/PFIP Financial Sector Assessment (2013) and updates provided by BPNG (2015). 12 Financial Inclusion and Financial Services in PNG provide deposit and credit services to members. There has been a steady reduction in the share of retail financial services held by S&Ls. In the 1970s there were over 120 S&Ls in PNG. There are currently 21 S&Ls. BPNG is seeking to strengthen and support S&Ls to ensure longer term sustainability (Bakani, 2014). Legislation is being revised to allow for the merger of S&Ls and for S&Ls to be able to borrow and access the payments system. S&L membership will also be broadened; the current ‘common bond’ requirement will be abolished. There are four commercial banks in Papua New Guinea, three of which offer retail financial services. The commercial banks are the largest providers of retail financial services in PNG and dominate the financial sector (Conroy, 2000). Relative to other regulated financial institutions in PNG, commercial banks have significant capacity which spans all aspects of financial services operations and management: governance, operational management, balance sheet and risk management, financial products and services and financial infrastructure. Each commercial bank has developed national distribution capacity, albeit primarily metropolitan, and offers a basic bank account (refer to Table 1). Whilst each of the three commercial banks has demonstrated a commitment to savings mobilisation through the provision of deposit and transaction services to poor and lower income customers, only one bank (BSP) has developed a branch and agent network of sufficient reach to achieve broad based geographic penetration. No commercial bank has developed a basic credit product targeted to financially excluded individuals, households and informal sector participants. Table 1 Commercial Bank Retail Banking Services ANZ Bank Bank South Pacific (BSP) Westpac Bank Network 12 branches and 42 ATMs 39 branches and 301 ATMs 17 branches and 37 ATMs Agents 72 GoMoney merchants 242 agents across PNG 100 in-store merchants Mobile banking Yes Yes Yes Rural banking Primarily metropolitan 41 dedicated BSP Rural locations Strategy is to focus on locations within two hours of an urban 12 area Basic bank account Go Money: Electronic only mobile Kundu Account: Branch and Choice Basic Account: Branch and banking service. Transaction electronic access. Transaction electronic access. Transaction and savings capability. No credit and savings capability. No credit and savings capability. No credit capability capability capability Financial literacy Yes Yes Yes programme 12    12   Key informant interview: Westpac Head of Retail Banking. 13 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 2 Microfinance Outreach 45 34 31 26 24 30 Financial Access Points 6 8 9 10 7 15 4 0 2 1 1 0 0 0 0 0 0 3 2 0 Enga Gulf Central NCD Milne Bay Jiwaka Eastern... Morobe New Ireland ENB WNB ARB Western... East Sepik West Sepik Northern Manus Madang Hela Chimbu Southern Region Highlands Region Momase Islands Region Region Source: BPNG (BPNG, 2015) There are five regulated microfinance institutions in PNG. Microfinance institutions provide deposit and credit services. Only one microfinance institution, MiBank (Nationwide Microbank) has nationwide representation (12 branches). As shown in Figure 2, Microfinance outreach is highly concentrated in a small number of provinces. 2.5. Accessing Retail Financial Services Access to retail financial services in PNG is bifurcated. In urban areas consumers have a range of institutions and products and services to select from. In the townships surveyed, consumers appear to have access to formal financial services. However, this is not the case with most rural communities where the principal issue is not choice or selection, but access to any form of formal retail financial services. PNG has one of the lowest levels of access point density globally. Financial Services Branches Of the 164 countries for which data is available for 2004, PNG ranked 138th. By contrast, of the 179 countries for which data is available for 2014, PNG ranked 169th. As shown in Figure 3, the average financial services branch density in PNG is 1.9 branches per 100,000 adults. By contrast, globally, financial services branch density 13 is 12.16 branches per 100,000 adults. The average branch density in lower-middle income countries is 7.8 per 100,000 adults (2004 = 5.06). In Pacific island countries (excluding PNG) the average branch density in is 18.9 branches per 100,000 adults. Branch density in PNG has remained constant between 2004 and 2013, whilst over the 13  http://data.worldbank.org/indicator/FB.CBK.BRCH.P5. 14 Financial Inclusion and Financial Services in PNG Figure 3 Branch Density per 100,000 Adults 20 15 10 5 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Lower Middle Income Pacific Island Countries Papua New Guinea World 14 Source: World Bank same period the average number of branches in other Pacific island countries grew by 14 20 percent and globally by 33 percent. It appears the situation in rural areas may have exacerbated in recent years, with an increasing concentration of branches in urban areas: ‘. . . licensed banks and savings and loan societies have been reducing their coverage in regional and rural areas over a period of decades’ (Conroy, 2000). Financial Services Agents Agents are often used to provide physical access points in locations in which it is not economical to open a branch. As examples, Bradesco (a Brazilian bank) operates an agent network with 24,500 locations nationwide (16.4 per 100,000 adults). In Kenya, a 15 network of almost 70,000 agents services ten banks in addition to Safaricom M-PESA 16 (268 per 100,000 adults). Tanzania has nearly 17,000 mobile money agents (57 per 100,000 adults). There are currently 26,000 agents (across multiple financial services 17 providers) in Peru (118 per 100,000 adults). There are currently approximately 771 agents in PNG providing services on behalf of regulated financial institutions (Anderson, et al., 2013). Overall agent density is modest (16.5 per 100,000 adults). 14  http://data.worldbank.org/indicator/FB.CBK.BRCH.P5. 15  http://www.afdb.org/en/news-and-events/article/fostering-financial-inclusion-with-mobile- banking-12125/. 16  http://www.cgap.org/blog/geography-cash-points-tanzania. 17   http://www.asbaweb.org/E-News/enews-40/fin/03fin.pdf (CGAP, 2015). 15 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 4 ATM Density per 100,000 Adults 40 35 30 25 20 15 10 5 0 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Lower Middle Income Pacific Island Countries Papua New Guinea World 18 Source: World Bank ATM and EFTPOS 18 Globally, ATM density has increased significantly over the past ten years. Global 19 average ATM density (2013) has now reached 34 ATMs per 100,000 adults. Similarly, ATM density has been growing quickly in Pacific island countries, having grown from 7.4 ATMs per 100,000 adults in 2004 to 22.0 ATMs per 100,000 adults in 2013. This represents a compound annual growth rate of 11.6 percent. The increase in ATM density in PNG has been slower than that of other Pacific island countries (refer to Figure 4). There are 398 ATMs deployed in PNG (BPNG, 2015). The compound annual growth rate is 8.0 percent. ATM density in PNG is 8.4 ATMs per 100,000 adults (2013). 20 Globally there continues to be strong growth in the number of EFTPOS terminals. There are approximately 13,500 EFTPOS terminals deployed in PNG (BPNG, 2015). There is no publically available country level data to enable a comparison of the density of EFTPOS terminals in PNG against other countries. Mobile Phone The 2009–2010 HIES found that overall, 49.1 percent of households in PNG owned a mobile phone. A slightly lower percentage of households in rural communities reported owning a mobile phone (42.5 percent). By contrast 89.2 percent of 18  http://data.worldbank.org/indicator/FB.CBK.BRCH.P5. 19  http://data.worldbank.org/indicator/FB.CBK.BRCH.P5. 20  http://www.thepaypers.com/cards/asia-pacific-gets-ahead-of-north-america-in-eftpos- terminals/757391-23. 16 Financial Inclusion and Financial Services in PNG households in urban communities reported owning a mobile phone. Given the work undertaken by mobile phone providers in PNG to extend rural mobile phone coverage, it is likely the percentage of the rural population owning or accessing a mobile phone has increased since the 2009–2010 HIES. BPNG has responded positively to the mobile banking opportunity. Mobile Network Operators have been given exemptions under Banks and Financial Institutions Act 2000 to conduct mobile phone banking. Levels of mobile phone banking penetration are currently not known, but are considered to be low relative to other financial services access channels. 17 3.  Overview of Morobe and Madang 3.1. Overview Morobe and Madang provinces are two adjoining provinces within what is termed the Momase (or sometimes the Mamose) region (an abbreviation of Morobe, Madang and Sepik one of the four regions of Papua New Guinea. The other regions are: the Highlands, New Guinea Islands and Papuan (or Southern) regions. PNG is geographically, economically and ethnically extremely diverse, from its communities in remote valleys in the Highlands provinces, to coastal and lowland communities on some of the country’s islands, coasts and accessible and inaccessible major valleys. Morobe and Madang are in some ways a microcosm of many of the diverse geographical and social characteristics of Papua New Guinea. It is impossible MAP 1  Papua New Guinea (highlighting Morobe and Madang Provinces) Source: Ezilon Inc © 18 Overview of Morobe and Madang for any province, or small selection of provinces to represent the entire country’s range of physical and human characteristics, nevertheless, Morobe and Madang do reflect some of that diversity. Both provinces span broad altitudinal ranges, from the country’s highest mountains, notably in the Bismarck, Finisterre and Sarawaget ranges, to the major lowland valleys of the Markham and Ramu rivers, and an array of wide and narrow coastal plains, with wet and drier climates, and large and smaller offshore and more remote islands. 3.2. Population Population Overview Momase has the second highest population by region, according to the 2011 National Census, with 1.9 million, comprising 26 percent of the country’s population. The largest region by population is the Highlands, recorded with 39 percent of the total population. Morobe province is the largest province in PNG by population (674,810 persons recorded in the 2011 National Census—9 percent of the National population), and the fifth largest by physical area, at 33,525 sq. km. Madang province has the fourth largest provincial population according to the 2011 Census (493,906—7 percent of PNG’s population), after Morobe, Eastern Highlands and Southern Highlands, and is also one of the larger provinces in the country by physical area. Figure 5  Population Density on Arable Land (per sq. km) 160 140 Head per km2 120 100 80 60 40 20 0 1990 2000 2011 Morobe Madang Source: NSO & PNGFA 19 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea MAP 2  Morobe and Madang Population Density Figure 6  Average Household Size—2011 7.4 5.9 5.4 5.5 5.3 Morobe Madang Nat Average Nat Urban Nat Rural Source: NSO 2000/2011 Census 20 Overview of Morobe and Madang MAP 2  Morobe and Madang Population Density (continued) 21 Source: Papua New Guinea Rural Development Handbook. Figure 7  Age Profile—Years of age (percent) 70 60 50 40 30 20 10 0 Under 15 15–64 64+ Morobe Madang 21 Source: NSO 2000/2011 Census 21   Hanson, Allen, Bourke & McCarthy (2001). 21 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea The populations of the two major urban centres, and the smaller towns of Morobe province, reflect the wide hinterland, with residents, including settlements, both from the local communities (notably Ahi in Lae) and from other parts of the two respective provinces and elsewhere across the Momase and Highlands regions. This provides both these centres with a diverse social and ethnic mix, as well as a wide range of household incomes. Papua New Guinea has a high proportion of its population under 20. The median age recorded in the 2011 National Census for the whole country was 21.4, in Morobe province it was 21.0 and for Madang province it was 19.0, with a national urban median of 21.9 and rural median of 21.3. Population Change According to the National Statistical Office, (based upon 2000 and 2011 Censuses), Papua New Guinea’s total population grew at an annual rate of 3.1 percent from 2000 to 2011, up from 2.9 percent during the 1990–2000 decade. 13 percent of PNG’s Figure 8  Age/Gender Breakdown 800,000 700,000 600,000 500,000 400,000 300,000 200,000 100,000 – Total Citizens Male Female Total Citizens under 18 Morobe Madang Source: NSO 2000/2011 Census Figure 9  Migrants 2011 (Percentage of Total Population) 45 40 35 30 25 20 15 10 5 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO 2000/2011 Census 22 Overview of Morobe and Madang Figure 10 Interprovincial Migration 2011 80000 70000 60000 50000 40000 30000 20000 10000 0 Inmigration Outmigration Net Migration Morobe Madang Source: NSO 2000/2011 Census total population was recorded as migrants in 2011 (i.e. not born where they were enumerated in the Census), with 40 percent of urban citizens recorded as migrants and 9 percent of rural citizens. Morobe, which includes the large urban centre of Lae, was above the national average and Madang province, with relatively small urban centres, was below the national average. In-migration comes from both rural areas within the respective provinces, and from other provinces, particularly heading for the main urban centres, like Lae and Madang, and commercial projects, in agriculture and mining. Major out-migration from Morobe and Madang provinces heads to the National Capital District particularly. 3.3. Economy and Livelihoods The provinces are diverse economically, with major mining and agricultural projects, from gold and nickel mining (notably in Bulolo District, Morobe, and Usino-Bundi in Madang) to oil palm, sugar, cattle and other livestock in the major Ramu and Markham valleys, coffee, vegetable and plantation forestry in the upland valleys, such as around Wau-Bulolo, and cocoa and copra, particularly on the major Madang island of Karkar. Commercial fisheries (and fish processing) are centred on the Madang and Lae ports and urban vicinities, tourism particularly in Madang and its lagoon hinterland, and logging also prevalent in certain coastal areas, particularly Madang province and parts of Morobe’s Siassi islands. Just over half the population of the two provinces recorded in 2011 was employed in subsistence agriculture (particularly women), just below the national average recorded, whereas about 12 percent of the Morobe population was recorded as engaged in wage employment, and just below 10 percent in Madang, both near the national average, and with higher percentages in paid employment amongst males. Both Morobe and Madang provinces recorded 70–80 percent of households growing food crops (around the national average); although the large urban population of Lae 23 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 11  Percentage of Employed Population in Subsistence Employment 2011 80 70 60 50 40 30 20 10 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO 2000/2011 Census Figure 12  Percentage of Employed Population in Wage Employment—2011 70 60 50 40 30 20 10 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO 2000/2011 Census pulled Morobe slightly below the average (only 20 percent of urban households are recorded as growing food crops). The record also shows that nearly 80 percent of Madang households participate in growing betel nut, significantly above the national average. Over 90 percent of households in the two provinces also recorded selling betel nut (and associated mustard), almost as many as growing, and food crops (somewhat lower the portion producing). Over 90 percent of households in the two provinces live in their own homes, roughly the national average, and these houses are largely made of traditional materials. 24 Overview of Morobe and Madang Figure 13 Household Profile—Livelihood and Home Ownership—percent 2011 120 100 80 60 40 20 0 Growing Growing Selling Selling Home Trade Food Crops Betel Nut Food Crops/ Betel Nut Ownership Dwellings Cooked Food Mustard Morobe Madang Nat Average Nat Urban Nat Rural Source: NSO 2000/2011 Census Lae and to a lesser extend Madang have become hubs not only of formal sector business but extensive informal sector trade, with a hinterland, particularly for Lae, stretching well beyond, as well as being major administrative and education centres, each with universities, teacher training and technical colleges and research institutes. 3.4. Infrastructure and Services Transport As across much of Papua New Guinea, transport infrastructure in Morobe and Madang provinces is rudimentary, with large portions of the provinces inaccessible by road, particularly during the wet season. The two provinces include some of the country’s major national highways, notably the Highlands (or Okuk) Highway up the Markham valley and the Wau-Bulolo road in Morobe province, and the Ramu-Madang Highway and north coast road to Bogia in Madang province. Nevertheless, these National Highways suffer from poor maintenance, and even become impassable at times. The provincial and local access roads into the districts suffer from much worse maintenance funding and, as with the rural airstrips, extensive stretches of these roads have closed down temporarily or permanently over the past two decades, 25 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea MAP 3  Access to Service Centre   leaving whole communities dependent upon pedestrian access only, and undermining access to markets and public services. As stated in the NEFC (National Economic and Fiscal Commission) report, ‘The Thin Blue Line’ (NEFC, 2014) “a large proportion of the road network has deteriorated to the point where ‘routine maintenance’ is not possible and major maintenance or rehabilitation is now necessary.” Yet, as that report highlighted, the cost of rehabilitation is much greater (“reconstruction of public assets can be 40–130 times more expensive than the annual cost of routine maintenance”), and that restoration of the entire network is unaffordable in the short–medium term. A World Bank sector study (2003) found that the severe decline in funding for maintenance, notably during the 1990s, resulted in a severe deterioration in the state of the country’s road assets and an accumulated backlog of work to be addressed. It found that only 49 percent of what was needed for road maintenance was provided in 1991, declining to 30 percent in 1994 and a mere 10 percent by 2001. Funding for 26 Overview of Morobe and Madang MAP 3  Access to Service Centre (continued)   22 Source: Papua New Guinea Rural Development Handbook infrastructure maintenance and restoration remained at only a fraction of the level required through the 2000s, only starting to improve modestly from the end of the 22 decades, along with improved revenue. Following the 1995 Organic Law on Provincial and Local Level Government the different roads and other infrastructure were meant to be classified as falling under National, Provincial and Local Level Government responsibility, respectively, with unconditional grants provided accordingly. Such classification of specific roads and airstrips, etc., and associated funding, did not materialise, and for at least a decade there was uncertainty, with much infrastructure left in a vacuum of responsibility for most non-national roads and airstrips. With the discontinuation of the Works Department in many provinces, there was also little or no human capacity or equipment at the District or LLG levels to undertake routine maintenance, and, as with 22   Hanson, Allen, Bourke & McCarthy (2001). 27 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 14  Nationwide Coverage by Digicel in 2011/2012 Current Coverage Planned Coverage 2012 Source: Digicel rural airstrips, which were closing at a steady rate, the limited maintenance which did occur was often left to ad-hoc community action and activity organised by service providers (such as mission air service providers). Based upon the 2005 Department of Works Road Asset Management System (RAMS) report, and their 2011 update, it has been calculated that an allocation of K15.2 million was required for basic maintenance of existing provincial, district and local level infrastructure (largely rural roads) in Morobe province and K12.6 million for Madang province, when unsealed provincial 23 roads were estimated to cost K10,500 per km to maintain. Only a fraction of these amounts have been available during these years and even smaller fraction of this level of funding has been provided for this purpose, although in recent years under the District Support Improvement Program (DSIP), K3 million has been provided per District for roads/transport. The planning and capacity have been largely deficient and much funding has been used on new facilities. However, in some districts, such as Bulolo and 23   2011 estimate—K15,000 PNG Department of Works. 28 Overview of Morobe and Madang Huon Gulf, substantial expenditure had been provided for restoration of certain roads and bridges in recent years. Telecommunications Whilst PNG’s main urban centres have been connected by telephone since the 1970s, using then state of the art microwave signals and mountain-top repeater stations, mobile phone access was only rolled out in the main urban centres in the early 2000s, with only very limited subscribers and access until mid-2007. The commencement of a competitor in 2007, and new investment in urban and rural services, saw the number of subscribers more than double within 3 weeks. Although still costly by world standards, access to mobile telephony and subsequently data services expanded considerably since 2007, unit costs fell and reliability and speed increased particularly in the urban centres, such as Lae and Madang, where 3G (and later 4G) services were introduced; in rural areas 2G services have remained predominant, although a program of upgrading has been ongoing. All the urban households in Morobe and Madang provinces have ready access to mobile-phone based telecommunications, and in 2014 most of the rural population had access, at least within a short walk from their villages. Only one of the current two operators of mobile telephone services provides any significant rural coverage. Electricity Power is only accessible in the main urban centres in Morobe and Madang provinces, in small townships and villages along the main highways and in the some district centres. The 2008 Household Income and Expenditure Survey (HIES) found that nationwide only 16.7 percent of the population had access to electricity from the grid in their households, 6.3 percent in rural areas, and 67.8 percent in urban areas, with a further 2.8 percent having access to privately generated power. In the Momase region (including Morobe and Madang provinces) the figure for access to the grid fell to 5.9 percent of the population, despite the presence of larger cities like Lae, and 1.8 percent private generation. Education School enrolment has been improving over the past decade, but enrolment, retention, standards and levels of literacy and numeracy remain low, especially in rural communities. Morobe and Madang provinces were both close to the national average in the 2011 National Census in terms of the percentage of citizens recording ever having been to school, at about 57–58 percent for the two provinces, as against approximately 56 percent nationwide. About 10 percent more males recorded having been to school in Morobe and nearer 7 percent in Madang, which was almost identical to the national average. Urban areas recorded almost 80 percent of citizens having attended school at some stage, as opposed to just over 50 percent in rural 29 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 15 School Attendance Attending School-2011 (% of Citizens) 50 40 30 20 10 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Ever Been to School 2011 (% of Citizens) 90 80 70 60 50 40 30 20 10 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO National Census 2011 Figure 16  Grade 10 Highest Grade (percent of Citizens) 40 35 30 25 20 15 10 5 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO National Census 2011 30 Overview of Morobe and Madang Figure 17 Literacy 2011 (percent of citizens) 100 90 80 70 60 50 40 30 20 10 0 Morobe Madang National Ave Urban Ave Rural Ave All Male Female Source: NSO National Census 2011 areas. 70 percent literacy was recorded amongst citizens in the Census in 2011 for Morobe province, and about 68 percent for Madang province, which was also about the national average, with rates amongst females about 4–5 percent lower than males. Urban areas again showed markedly higher literacy rates, at almost 80 percent average, than rural areas, at somewhat above 60 percent. Health Papua New Guinea health indicators are poor and although there have been some improvements, for example in reduced malarial incidence over the past decade, key indicators (including health Millennium Development Goals for maternal and child health and mortality rates) remain extremely unsatisfactory. For example infant (under 1 year of age) mortality rates were recorded as reducing from 69.3 deaths per 24 1,000 live births, to 56.7 deaths between 1996 and 2006. The equivalent figures for MOMASE region were 76/1,000 in 1996 and 55/1,000 in 2006. Urban infant mortality rates are 32 per 1,000 live births, but rural rates of 62/1,000, and major variations relate to levels of education, than with infant mortality rates more than twice as great for those with no education, than those with grade 7 or more. Health is determined by multiple factors, including access to information, women’s education and empowerment, access to economic opportunities, as well as natural environment and access to quality health services. In much of Papua New Guinea health services have been recorded as having deteriorated over the past decade, in both range of the services provided and quality (Howes et al., 2014). The cost of health service provision varies considerably from one 24   Papua New Guinea Demographic and Health Survey 2006, National Statistical Office, October 2009. 31 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea part of the country to another, as determined particularly by remoteness. The largest portion of costs (approx. 84 percent) was calculated by NEFC to be needed at the District level. Morobe was estimated by NEFC to require approximately K11 million for health sector costs, and Madang around K 9 million, with a substantial portion of that explicitly for patient transfers. Based upon the Department of Health’s policy on patient transfers and the NEFC’s costs of services study (reflecting the portion of the population living in relatively isolated locations and other factors), Morobe, with both its large population and large rural population requires K3.6 million for patient transfers alone, and Madang requires K2.6 million: the highest transfer costs by province. This level of funding was not made available. According to the report ‘Below the Glass Floor’ (World Bank, 2013a) frontline spending in Morobe province was ‘low’ compared to what was required (with consistently low levels of funding for health services, including water supply and sanitation, notably in 2009/10), and ‘medium’ for Madang province (where funding was increased markedly in 2009 and 2010). Please refer to Appendix 1 for a brief overview of the Districts in Morobe and Madang in which field-work was undertaken. 3.5. Financial Services There is good access to formal financial services access points in urban and township locations. Commercial banking and finance company services are available in the two provincial capitals (Lae and Madang). Retail banking services, including ATM and EFTPOS services are available at multiple locations (refer to Table 2) in Lae and Madang and also in the townships of Bulolo and Finschafen. ATMs are principally co- located with bank branches. There are, however, several stand-alone ATMs in Lae and 25 Madang (on the Divine World university campus for example). 25 Table 2  Access to Financial Services Morobe and Madang Banks Micro- ANZ BSP WBC Total S&Ls finance Branches Urban 3 4 2 9 2 2 Township 0 1 0 1 0 2 ATMs Urban 5 16 6 27 NA NA Township 0 1 0 1 NA NA Agents/ Urban 11 4 ND NA NA In store Rural ND 23 ND NA NA 25  Information sourced from financial services provider websites and BPNG. 32 Overview of Morobe and Madang Savings and Loan and microfinance institutions also have outlets in the provinces. Similar to the commercial banks and finance companies, S&Ls and microfinance organisations are located in urban and township locations. Post PNG has six outlets in Morobe and Madang, and also in urban or township locations. Overall, for 1st Quarter 2015, Momase had a total of 2,579 formal financial services access points (including EFTPOS). Morobe province had 1,586 and Madang 475. The number of financial services access points has increased by 38.7 percent in Morobe 26 and 18.31 percent in Madang over the last three year (2012–2015). In addition to accessibility to financial services access points, households in urban and township communities in Morobe and Madang have access to a full range of formal financial services. Savings, long term savings and investment services and credit services (including asset finance) are offered by multiple financial services providers. Levels of competition have not been examined by the present study. However, a range of products, within product categories and price points, can be selected. Consumers in urban location appear to have both access and choice. The situation in rural communities in both Morobe and Madang is very different to that of urban and township communities. Only one bank, BSP, has a rural agent network. The number of rural agents, relative to the rural population is, however small. A very significant proportion of the rural population in Morobe and Madang effectively have no, or very limited access to formal financial services. In addition, the range of financial services available to rural communities is very limited. Products and services are limited to those which can be offered by agents. Complex products and products which require financial advice prior to purchase (for example credit and long term savings products) have very limited availability. Overall, it appears most consumers in rural location in Morobe and Madang have neither access to formal financial services or a set of financial products and services from which to select an appropriate product. 26   Source: Bank of PNG. 33 4.  Financial Inclusion in Morobe and Madang 4.1.  Facilitators of Financial Inclusion In addition to examining use of financial services, respondents’ confidence in communicating in English, both oral and written, and access to and ability to use a mobile phone was also examined. Confidence with Communicating in English The ability to communicate in using English facilitates financial inclusion. The ability to communicate in English is not a pre-requisite for financial inclusion. It is likely many counter interactions will be undertaken in Tok Pisin. However, most documents are written in English. This includes account opening and contract documents, many brochures and EFTPOS and ATM receipts. The ability to communicate in English, in particular the ability to be able to read documents written in English facilitates financial inclusion. Nearly 60 percent of respondents stated they could not communicate in English. Responses tended to bifurcate. Most respondents stated that either they could not communicate in English at all, or they could communicate in English both verbally and using written media (read and/or write). Very few respondents stated they could only speak English. Overall men were more likely than women to state they were able to communicate in English (46.8 percent of men compared to 26.1 percent of women). There were significant location based differences in respondents’ confidence in communicating in English. Eighty-two percent (82 percent) of urban respondents and 56 percent of township respondents stated they could communicate in English, compared to 29 percent of rural respondents. All men and 75 percent of women living in urban communities stated they could speak English. By contrast only 38 percent of men and 23 percent of women living in rural communities stated they could speak English. The location and gender differences in respondents’ confidence in communicating in English are summarised in Figure 18. Use of Mobile Phones The lack of financial services infrastructure in many rural communities suggests the extension of financial services to rural communities will require the use of mobile 34 Financial Inclusion in Morobe and Madang Figure 18  Confidence in Communicating in English 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Male Female Male Female Male Female Urban Township Rural Cannot Communicate in English Speak in English Only Speak and Read and/or Write in English phone telephony, to both enable agents and merchants to connect to the payments system and to enable financial services customers to access their accounts. Access to a mobile phone and the ability to use mobile phone text functions therefore facilitates financial inclusion. There are significant differences in the level of mobile phone ownership and ability to use mobile phone functionality between urban and rural communities and men and women. As discussed above, the 2009–2010 HIES found that, overall, 49.1 percent of households in PNG owned a mobile phone. A slightly lower percentage of households in rural communities reported owning a mobile phone (42.5 percent). By contrast 89.2 percent of households in urban communities reported owning a mobile phone. Overall it is considered likely the level of ownership or access to mobile phones in rural communities in PNG will have increased since the last HIES. The present study found similar levels of mobile phone ownership to those reported in the 2009–2010 HIES. As shown in Table 3, in urban and township communities, access to a mobile phone is high. Most respondents who owned or had access to a mobile phone were also able to use the phone to text as well as using the phone for calls. By contrast, levels of mobile phone access in rural areas were low, particularly by women. This may be due, at least in part, to several of the CUs sampled having no mobile phone access. Mobile phone capability was also more limited, again particularly by women. The combination of access and capability suggests that there is a high capacity for mobile phone based financial services in urban communities. At present, however, the capacity to extend mobile phone banking into rural communities may be constrained by access and capability issues, in particular for women. It appears that, without capacity development, approximately 30 percent of men and 10 percent of women living in rural communities in Morobe and Madang may currently be able to use a mobile banking service. 35 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 3 Ownership and Use of Mobile Phones Urban Township Rural Male FEMale Male FEMale Male FEMale Own/access a mobile phone 94.40% 83.60% 96.60% 76.10% 48.40% 21.80% Make calls and texts 85.90% 87.90% 80.50% 36.10% 61.20% 44.90% Capacity for mobile phone banking 81.20% 73.50% 77.8% 27.5% 29.60% 9.80% The remoteness of the rural community from an urban area does not appear to be a principal driver in mobile phone access and usage. Seventy three percent (73 percent) of respondents living within an hour of an urban area stated they owned or could access a mobile phone. This reduced to 40 percent for respondents living 1–3 hours, and 31.6 percent for respondents living 3–9 hours from an urban area. However 56.7 percent of respondents living more than 9 hours from an urban area stated they owned or had access to a mobile phone. It is, of course, likely the respondents’ ownership/access to a mobile phone is related to the presence in the area of a mobile phone tower. 4.2. Use of Formal Financial Services This section of the report examines the number of financial products owned by respondents, and the range of products owned across the following categories: Savings, long term savings and credit. Respondents’ payment and remittance activity is also overviewed. Number of Financial Products Owned The number of financial products (both formal and informal) owned by respondents, varied significantly by gender, location and source of livelihood. As shown in Table 4, in rural communities 60.2 percent of men and 81 percent of women reported owning no financial product; by contrast, in urban communities men reported owning 3.4 financial products on average and women own only 1.3 financial products. There were significant differences in financial product ownership by principal source of livelihood. Perhaps not surprisingly, those in formal employment were significantly more likely to report owning financial products than those in informal employment. However, even within livelihood categories, men were more likely to report product ownership than women. Nearly 93 percent of men employed in the formal sector reported owning at least one financial product, compared to 66 percent of women. Whilst information on a specific occupation was not collected, it is possible the higher level of account ownership by men working in the formal sector may relate to employment in organisations which pay wages by credit to a bank account, rather than in cash. 36 Financial Inclusion in Morobe and Madang 26 Table 4 Number of Financial Products Owned Urban Township Male Female Male Female Rural Informal Informal Informal Informal Sector/ Sector/ Sector/ Sector/ Formal Self- Formal Self- Formal Self- Formal Self- Sector employed Sector employed Sector employed Sector employed Male Female 0 7.30% 22.30% 36.20% 43.90% 10.30% 25.00% 29.70% 66.40% 60.20% 81.00% 1 14.00% 21.30% 23.00% 35.10% 9.70% 50.00% 27.70% 16.80% 18.80% 9.20% 2 7.30% 10.20% 8.20% 10.00% 14.80% 0.00% 0.00% 16.80% 8.40% 5.10% 3> 71.40% 46.20% 32.70% 11.00% 65.20% 25.00% 42.60% 0.00% 12.60% 4.70% Account ownership by urban and township respondents who were self-employed or working in the informal sector was lower than for respondents working in the formal sector (refer to Table 4). Levels of financial inclusion were, however, significantly higher than those reported by respondents employed in the informal sector in rural communities. Fifty-six percent (56 percent) of women and nearly 78 percent of men reported owning at least one financial product. Overall, whilst there continues to be a significant number of urban dwellers who do not have at least one financial product, in particular women, the large scale financial exclusion appears to be far more 26 pervasive in rural communities than in urban communities.   < > Overview of Financial Products Owned Financial products were grouped by category. The savings category included debit card based savings and transaction products, and passbook based products from banks, S&Ls and microfinance providers. The long term savings category included term deposits, provident/superannuation, unit trust/shares and life insurance. Credit included any form of formal credit and Protection covered house/contents or car insurance. Overall, as shown in Table 5, those employed in the formal sector were more likely to own financial services in each of the product categories, in particular savings and long term savings. Men were more likely to report product ownership in each product category than women. Overall men were twice as likely as women to report product ownership. 26  Responses by urban and township respondents are presented by gender and by livelihood. Rural respondents are presented by gender only as 88 percent of respondents were employed in the informal sector/self-employed. 37 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 5  Financial Product Ownership by Category Urban Township Rural Livelihood Informal Sector/ Formal Self- Male Female Male Female Male Female Sector Employed Savings 68% 38% 83.60% 24.30% 21% 9% 62.30% 13.40% Long term savings 56% 28% 45.40% 17.10% 10% 5% 59.70% 3.50% Credit 50% 16% 13.00% 6.80% 6% 1% 25.50% 3.30% Protection 8% 8% 0% 0% 2% 0% 2.00% 1.50% Households in urban communities have access to a broad range of financial services from a range of financial services providers. Evidence from Morobe and Madang suggests that, whilst formal sector employment provides a pathway to ownership of a savings/transaction account and long term savings, 50–75 percent of urban respondents deriving informal sector income as their principal form of income were also likely to own at least a savings account. The majority of rural households, however, have little or no access to the formal financial system and own few financial products. Payments and Remittances This section examines modalities used for payment for goods and services purchased and the receipt of income, and also remittance modalities used by households, both outward and inward remittances. Payments As shown in Table 6, nearly all households reported incurring expenditure on day- 27 day items, community and religious obligations and expenditure on education. Not surprisingly few rural households reported expenditure for services (70 percent of urban household reported having mains electricity, compared to 7 percent of rural 28 households) and were less likely to incur expenditure for rent/lease payments or taxes. Overall, urban households and households in townships were more financially active and both more likely to incur expenditure across a wider range of expenditure categories and to be more likely to incur expenditure within each category. 27  Any expenditure on education, not necessarily school fees. 28  Refer to Appendix 4 Household Overview. 38 Financial Inclusion in Morobe and Madang Table 6 Households Incurring Expenditure by Type Urban Township Rural Day-day items such as food or transport 97.40% 100% 98.40% Bills (e.g. electricity or water) 61.40% 91.70% 9.60% School or university fees 75.70% 78.30% 63.70% Loan repayments 37.50% 35% 12.00% Rent/lease payments 36.80% 21.50% 10.30% Community/religious donations 91.90% 93% 80.10% 29 Levies or taxes 36.30% 38% 19.40% As shown in Table 7, the most common mode of payment for day-day expenditures and bills, irrespective of whether the household had an account, was cash. Nevertheless, approximately 21 percent of urban households and 12 percent of rural households that reported owning a card-based account also reported paying for day-day items using electronic payment. This suggests increased EFTPOS and ATM penetration is resulting in 29 a change to payment behaviour irrespective of location, shifting payments away from Table 7  Payment Modality for Expenses Incurred by the Household Has Account Does not have Account Urban Township Rural Urban Township Rural Day-day items such Cash 79.2% 95.5% 88.0% 95.2% 97.2% 99.1% as food or transport Electronic—Bank 20.8% 4.5% 12.0% 4.8% 2.8% 0.7% Bills Cash 61.3% 81.5% 93.0% 88.4% 90.3% 100.0% Electronic—Bank 27.7% 18.5% 0.0% 11.6% 9.7% 0.0% Mobile phone 11.1% 0.0% 7.0% 0.0% 0.0% 0.0% School or Cash 31.4% 74.0% 88.6% 49.2% 72.7% 91.0% university fees Electronic—Bank 68.6% 26.0% 11.4% 50.8% 27.3% 8.2% Loan repayments Cash 17.1% 17.8% 41.3% 31.7% 0.0% 47.5% Electronic—Bank 78.2% 82.2% 58.7% 68.3% 100.0% 52.5% Rent/lease Cash 38.3% 39.5% 80.0% 84.2% 100.0% 92.5% payments Electronic—Bank 61.7% 60.5% 12.2% 15.8% 0.0% 7.5% Community/ Cash 100.0% 100.0% 86.2% 100.0% 100.0% 96.2% religious donations In-kind/non-cash 0.0% 0.0% 13.8% 0.0% 0.0% 3.8% 29  Includes formal taxation and community levies. 39 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea cash. The shift to electronic media is also evident in the use of mobile phone payment. In both urban and rural communities, households with a bank account were more likely to use the mobile phone to pay for services (7 percent of rural households and 11 percent of urban households). The use of electronic (bank-bank) payment for school/university fees, loan repayments and rent/lease payments was common, irrespective of whether the household had an account. The reasons for this are not immediately obvious and follow-up research may be warranted (or perhaps included in further deployment of the financial capability survey) to determine how households who do not have a bank account make electronic payments (other than through the Post Office or a wire service). It is also possible there are language issues. As discussed in Chapter One, translation of financial constructs from English to Tok Pisin can be challenging due to the lack of construct equivalence. Receipts Patterns of income receipt varied between households in urban communities and those in rural communities. As shown in Table 8, urban and township households were significantly more likely than rural households to report receipt of wages (both formal and informal wages). Most households reported receipt of payments for sales. In urban 30 households this appears to be primarily secondary income. Urban households were also more likely to report receipt of rent/lease and interest income. Payment for sales was primarily cash based, irrespective of location or whether the respondent had a bank account (refer to Table 9). A similar issue emerges in respect to receipt of wage/salary income (and also rental income in urban households) as discussed above for electronic payments by respondents who reported not owning a bank account. Reported levels of receipt of wages/salary to an account were high, even for respondents who reported they did not own an account. There are Table 8 Households Receiving Receipts by Category Urban Township Rural Wages/salary 59.70% 62.00% 15.00% Payment for sales 75.30% 83.00% 95.00% Royalty payments 7.50% 10.00% 6.40% Rent/lease payments 22.60% 15.00% 8.40% Interest 16.20% 18.30% 8.50% 30  Refer to Appendix 4 Household Overview. 40 Financial Inclusion in Morobe and Madang Table 9 Receipt Type for Income Received by the Household Urban Township Rural Does Does Does Has not have Has not have Has not have Account Account Account Account Account Account Wages or salary Cash 21.60% 15.10% 7.80% 17.60% 30.50% 64.10% Electronic—Bank 78.40% 84.90% 92.20% 82.40% 69.50% 35.90% Payment for sales Cash 100.00% 100.00% 100.00% 95.3%* 97.40% 99.80% Electronic—Bank 0.00% 0.00% 0.00% 0% 2.60% 0.20% Royalty payments Cash 50.00% 100.00% 100.00% NA 80.80% 100.00% Electronic—Bank 50.00% 0.00% 0.00% NA 19.20% 0.00% Rent/lease Cash 60.50% 68.80% 50.00% 20.50% 77.30% 100.00% payments Electronic—Bank 39.50% 31.20% 50.00% 79.50% 10.80% 0.00% Interest Cash 32.30% 100.00% 28.30% NA 40.70% 87.90% Electronic—Bank 67.70% 0.00% 71.70% NA 48.80% 0.00% In-kind/non-cash 0.00% 0.00% 0.00% NA 10.50% 12.10% *in-kind 4.7% instances where one member of the household reported receiving wages or salary by electronic transfer, but it was another member of the household who reported having a bank account. Further investigation is required and there may be value in adding questions to the national survey to explore payment and receipt using a third-party bank account. It is possible income received by household members who do not have a bank account is directed to a bank account held by another household member (within the context of this study typically the partner). It is also possible that some households bank accounts, irrespective of ‘ownership’, are managed or controlled by one member of the household who may receive income on behalf of other members of the household. The low level of cash receipt of wages/salary by township houses may be due to sampling (in particular Hospital and Forestry employment). Remittances Remittance activity (sending or receiving funds to a person in another location) was more common in urban households than rural households. As shown in Figure 19, nearly 80 percent of urban respondents stated they sent remittances and 60 percent that they received remittances. The most common method of sending or receiving a remittance in urban communities was by bank transfer. Interestingly approximately 5 percent of remittances were sent/received using the mobile phone. The use of the mail or personal delivery was also common. The PNG Financial Diaries Study (Stuart, 41 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 19 Remittances Sent and Received 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Send Receive Send Receive Send Receive Urban Township Rural Mobile Phone Post Office Bank Mail/Personal Delivery Noggle, & Sibley, 2014) (also sponsored by BPNG) mapped domestic remittances from participants based in or proximate to Port Moresby, Goroka and Kimbe. The remittance transactions captured by the financial diaries study indicated participants made remittance transfers to a wide spread of locations across PNG. If the pattern found by the financial diaries study is replicated in Morobe and Madang, it is possible that remittance transfers to a large number of rural communities may need to be effected by mail or personal delivery, simply because there is no other way of effecting the remittance. The density of bank branches in rural communities is low across PNG, and 31 Post PNG estimates 60 percent of the PNG population cannot access a postal outlet. This may also explain the relatively low level of SMK transfers cited by respondents. Levels of remittance activity in rural communities, both sending and receiving funds, were approximately half those of urban communities. The dominant form of remittance was mail or personal delivery. As stated above, it is likely this is due to the lack of rural financial services and postal services infrastructure in many rural communities. Savings How Household Cash Is Kept Safe In general terms, households that had a bank account tended to use the account for safekeeping of cash, whilst households which did not have a bank account tended to keep money hidden or in a locked box (or similar). Thirty four percent (34 percent) of rural households that had a bank account also kept money hidden. This may be due to the distance required to travel to an access point to lodge or withdraw funds. 31   Key Informant Interview Post PNG Operations manager. 42 Financial Inclusion in Morobe and Madang Savings Accounts Globally, 50–60 percent of adults in developing economies report owning an account at a bank or another type of financial institution (Demirguc-Kunt, Klapper, Singer, & Van Oudheusden, 2015). As shown in Table 10, in Morobe and Madang, whilst men are more likely to report savings account ownership than women, overall more than 40 percent of adults living in urban or township areas reported owning a savings account, primarily a savings account with EFTPOS and ATM capability. In rural areas less than 10 percent of the adult population reported owning an account, whether card based or passbook based. The correlation between ownership of savings account and distance from the nearest bank branch is significant (r = .229, p < .001). The average distance to a bank branch for urban or township residents is estimated to be less than 5km, for rural residents it is estimated to be 38km (with a significant number of households living more than 38km from the nearest bank branch). The correlation between savings account ownership and formal employment is also significant (r = .263, p < .001). The two factors are, of course related given the concentration of formal employment in urban areas. The relationship between account ownership and distance to a bank branch is shown graphically in Figure 20. Table 10 Savings Account Ownership Urban Township Rural Male FEMale Male FEMale Male FEMale Savings/cheque 55% 28% 52% 24% 10% 8% (debit card) Savings (passbook) 32% 13% 49% 4% 14% 1% Figure 20  Bank Account Ownership and Distance to Nearest Branch 45.% 45 40.% 40 Distance to Bank (KM) Has a Bank Account 35.% 35 30.% 30 25.% 25 20.% 20 15.% 15 10.% 10 5.% 5 0.% 0 Urban Township Rural Has a Bank Account Distance to Bank 43 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 21  Propensity to Save 90.0% 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Try to Save Money Try to Save Try to Have Some Provision for for the Future Money Regularly Emergencies/Unexpected Expenses Does Not Have Savings Account Formal Sector Does Not Have Savings Account Informal Sector/Self-Employed Has Savings Account Formal Sector Has Savings Account Informal Sector/Self-Employed There is an overlap in the type of savings account owned, 21 percent of debit card based savings account owners also reported owning a passbook account. Dual account ownership was reported for both urban and rural households, and for formal and informal sector livelihoods. Commercial banks in PNG no longer offer passbook based accounts, whilst other financial institutions (in particular microfinance banks) continue to offer passbook based savings account. This suggests up to one in five adults who have an account at a commercial bank may also have an account at another regulated financial institution. There is tentative evidence that ownership of a savings account increases the propensity to save. As shown in Figure 21, irrespective of whether income is derived from (regular) formal sector wages or salary or (more intermittent) informal sector employment or self-employment, households that have a savings account are more likely to endeavour to save money regularly for the future and to have a provision for emergencies or unexpected expenses. Long Term Savings and Investments Ownership of long term savings products, in particular provident fund membership, is associated with employment in the formal sector. Sixty percent (60 percent) of adults employed in the formal sector reported having some form of long term savings 44 Financial Inclusion in Morobe and Madang Table 11 Long Term Savings Ownership Urban Township Rural Male FEMale Male FEMale Male FEMale Fixed term deposit 11% 0% 0% 0% 2% 1% Unit trust/shares 14% 6% 6% 0% 4% 1% Life insurance 25% 8% 0% 0% 2% 1% Provident/ 44% 25% 45% 17% 7% 4% superannuation product, compared to 3.5 percent of adults employed in the informal sector. Whilst respondents cited ownership of a range of long term products, issues with translation to Tok Pisin may result in the blurring of categories, in particular between provident/ superannuation, life insurance and unit trusts. Approximately 50–55 percent of formal sector employees reported having a provident or superannuation account. The reasons why 45–50 percent of adults who reported formal sector employment did not also report membership of a provident/ superannuation fund are not known. It is possible adults who did not report provident fund membership may have been primarily casual wage workers, or employed in small 32 firms. As shown in Table 11, the gender-based difference in employment patterns results in men being significantly more likely to report owning long term savings and investment products than women, irrespective of location. Credit The Global Findex (Demirguc-Kunt, et al., 2015) found approximately 9 percent of adults in developing countries reported borrowing from a formal financial institution. The present study found approximately 7 percent of households reported some form of formal credit obligation. Urban households were significantly more likely to report a formal loan than rural households (32 percent of urban households, compared to 3 percent of rural households). Within the context of urban households, formal credit obligations are more common in households in which the principal source of income was formal sector wages/salary (42 percent of households) than informal sector income (24 percent of households). Whilst it is possible that one in five urban households has a secured loan and approximately one in four households has a business loan (refer to Table 12), the levels 32   Provident contribution is only mandatory for employees in firms with at least 10 employees. 45 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 12  Formal Credit Obligations Urban Township Rural Male FEMale Male FEMale Male FEMale Commercial/development loan 27% 7% 0% 7% 3% 1% Secured personal/house loan 20% 6% 13% 0% 1% 1% Unsecured personal/house loan 11% 3% 6% 0% 2% 0% Credit card 8% 3% 0% 7% 4% 0% of reported urban formal sector credit should be regarded as tentative, in particular when compared to levels of insurance policy ownership (which would usually be associated with security commitments for formal loans). Once the national study has been completed, levels of reported formal sector borrowing should be correlated to levels of consumer lending by banks, savings and loans and microfinance institutions (and potentially finance companies). Protection Levels of home and car insurance were measured by the present study and very low levels of insurance were reported. Less than two percent (2 percent) of households in Morobe and Madang reported owning house or motor vehicle insurance. Approximately 8 percent of urban households reported having an insurance policy, compared to 1 percent of rural households. The level of insurance is significantly lower than the reported level of secured borrowing (and underscores caution in respect to reported levels of secured borrowing). There would appear to be a lot of work to be done in order to progress the use of 33 insurance generally in PNG. Anecdotal evidence indicates the issue of very low levels of insurance may be pervasive in PNG and may not be limited to insurance over domestic property. 4.3. Barriers to Financial Services Usage Access to Financial Services It is evident that many households in Morobe and Madang, in particular rural households, have limited engagement with the formal financial sector. The estimate of 85 percent of the adult population being excluded from the formal financial sector 33   Key Informant Interview Insurance Commissioner. 46 Financial Inclusion in Morobe and Madang appears to be generally valid for Morobe and Madang. Banks are the only participants in the payments system and therefore debit card issuers. The present study has found levels of debit card based savings account ownership to be 15 percent. Overall 22 percent of the population of Morobe and Madang reported owning some form of savings account. Whether a narrow measure or broad measure of account ownership is used, the figure masks a very major difference in levels of financial exclusion across Morobe and Madang. Urban households, whether deriving income from formal sector employment or participation in the informal sector, exhibit levels of financial inclusion (as measured by account ownership) similar to or greater than those typically found in developing countries. It is to be expected households deriving income from formal sector employment will be more likely to own a bank account and to have some form of long term savings (primarily provident). This appears to be the case for households deriving formal sector income in Morobe and Madang, irrespective of whether the household is located in an urban or rural location. However, formal sector employment in rural communities is limited. Overall, levels of financial inclusion in rural communities are very low. The principal issue appears to be lack of rural financial services infrastructure. Most rural households simply cannot access the formal financial system. The difference in the use of formal financial services between urban and rural communities is pervasive across all product categories. Households in rural communities that have a bank account are more likely to use cash for payments and less likely to remit or receive money electronically than urban households, are significantly less likely to keep cash safe in an account, are less likely to have long term savings and less likely to have a formal loan. Gender Women in PNG appear to be significantly more likely to be financially excluded than men and, if women do have an engagement with the formal financial sector, the scope of the engagement is generally more limited than for men. The gender gap in respect to the ownership of financial services appears to be significantly larger in Morobe and Madang than is typical for countries other than PNG. Globally 58 percent of women reported owning a bank account, compared with 65 percent of men, a 7 percentage 34 point difference. Levels of financial inclusion (as measured by ownership of a bank account) are lower in Morobe and Madang than the global average. However, as discussed above, in urban households levels of account ownership are significantly 34  http://www.wsj.com/articles/financial-inclusion-gender-gap-persists-for-bank-account- holders-1429128001. 47 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea higher than in rural communities. Sixty eight percent (68 percent) of men and 38 percent of women in urban households reported owning some form of savings account. Levels of financial inclusion, across all product categories are lower for women than for men. In part this can be explained by the greater percentage of women earning informal income. However, even for women earning formal sector income, levels of participation in the formal financial sector are lower than those of men. The reasons for this cannot be attributed solely to access. Most households in urban areas need to travel less than 5km to access financial services. The present study has sought to develop an understanding of levels of financial inclusion and product usage. The study has not sought to explain why there is such a pervasive and persistent difference in levels of financial inclusion between men and women, and the range of factors which appear to exacerbate gender difference in financial inclusion in urban communities. The issue is significant and it is possible that, as levels of financial inclusion increase across PNG generally, the gender gap will continue to widen. Further investigation is required to understand gendered drivers of financial exclusion and to facilitate the development of programs to bring greater numbers of women into the formal financial system. Nevertheless, reported urban levels of account ownership by men suggest financial inclusion programmes in Morobe and Madang should focus on women living in urban communities and focus on both men and women living in rural communities. Affordability The cost of financial services is often cited as one of the principal reasons for financial exclusion in PNG, in particular financial exclusion by poor and low income households. It is not uncommon, for example, when conducting focus groups to be advised that a member of the focus group either has opened a bank account, or knows someone who has opened a bank account and deposited a sum of money, and then checked the balance sometime later to find the account had been closed as fees had extinguished 35 the balance. The veracity of these statements has never been tested. Typical financial account transaction patterns for poor households in PNG are not known. The PNG Financial Diaries Study suggested 2–3 account transactions per month (one deposit and two withdrawals). However levels of account ownership by participants in the study were very low. In an effort to develop some understanding of the likelihood of account affordability, two transaction scenarios were modelled against the poverty line. It was assumed a household living on the poverty line has a basic bank account and 36 undertakes the following transactions: 35  This occurred during the development of the Financial Capability Survey instrument. 36  It is stressed the model is tentative and does not reflect actual account usage patterns. 48 Financial Inclusion in Morobe and Madang Scenario 1— Scenario 2— High transaction Low transaction Teller deposit 1 per week 1 per fortnight ATM withdrawal 2 per week 1 per week Bill payment (electronic or mobile) 3 per month NA Scenario 1: Regular receipt of funds, savings and payments Scenario 2: Savings only The poverty line for PNG varies between NCD (Port Moresby) and the rest of PNG. The poverty line in NCD is estimated to be approximately PGK3500pa. The poverty line for the rest of PNG is estimated to be approximately PGK1700pa (World Bank, 2012). The annual cost of financial services (the sum of account keeping and transaction fees) was modelled for the highest cost basic commercial bank account and lowest cost basic 37 commercial bank account currently available in Morobe and Madang. Annual account costs are shown in Table 13. Account keeping costs could range between 3 percent and 22 percent of household income at poverty line if an account is used for savings only and 8 percent and 40 percent of household income at poverty line if an account is used for regular receipt of funds, savings and bill payments. The model suggests the affordability of financial services may be a function of access to a range of financial services. If the lowest cost option is available to poor and low income households, it is likely a significant number of households would be able to afford a basic bank account for savings. The individual income estimates provided by participants in the present study suggest higher levels of affordability. Table 13  Estimate of Bank Account Costs as PERCENT Income for a Household on the Poverty Line Scenario 1— Scenario 2— High Transaction Low Transaction Low Cost High Cost Low Cost High Cost Account cost $PGK Per year $288 $688 $117 $374 Per week $6 $13 $2 $7 Account cost as NCD 8% 20% 3% 11% percent of household Rest of PNG 17% 40% 7% 22% income at the poverty line 37  Only basic accounts from commercial banks were modelled. Costs would be lower for microfinance or S&Ls. 49 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea 4.4. Use of Informal Financial Services Understanding of the pervasiveness and use of informal financial services in PNG, both savings and credit, is limited. The financial capability survey collected information in respect to product ownership. Information was not collected in respect to transaction frequency, cost and functionality. The pervasiveness of informal financial services, particularly in rural areas where formal financial services are commonly not available, suggests further research may be warranted. Savings Evidence from the present study suggests informal savings is associated with employment (refer to Table 14). Respondents were asked if they participated in a Sunday-Sunday and, in addition to being asked about provident/superannuation savings, respondents were asked if they saved using a scheme provided by their 38 employer. A previous study had found employees working on a plantation used an employer provided savings scheme to provide for their children’s education. As has been stated in other sections of this report, there are construct issues translating financial terms from English to Tok Pisin. Nevertheless, 34 percent of respondents who stated they contributed to provident/superannuation also stated they saved with their employer. Even allowing for some misinterpretation between employer provided savings, employment based ‘Sunday-Sunday’ schemes and provident/superannuation contributions, informal savings appears to be closely related to formal sector employment, irrespective of the modality. Credit Global evidence indicates adults in developing countries are more likely to borrow from family and friends than other sources. Demirguc-Kunt and colleagues (2015), reviewing Global Findex data found borrowing from family and friends to be the most common source of finance for people in all countries other than high income Table 14  Informal Sector Savings Informal Sector Formal Sector Savings with an employer 1.70% 28.80% Sunday-Sunday savings 3.00%  8.30% 38  The financial competency study in Port Moresby, Mekeo and Galley Reach. 50 Financial Inclusion in Morobe and Madang countries. Findex data indicates 29 percent of adults reported borrowing from family or friends. In several regions more people reported borrowing from a store than reported borrowing from a financial institution. Less than 5 percent of adults around the world reported borrowing from a private informal lender (Demirguc-Kunt, et al., 2015). As shown in Table 15, levels of borrowing from money lenders found in Morobe and Madang are similar to levels found in countries other than PNG. The characteristics of money lending in PNG are not well understood. Information tends to be anecdotal and is frequently pejorative, equating money lending with loan sharking. Evidence from the focus groups undertaken as part of the development of the financial capability survey instrument indicated that, at least in some rural areas, money lending may not be loan sharking but may be means of augmenting household income as well as being a source of funding for the household: adults in households with surplus liquidity lent the money to households with a liquidity shortfall. As shown in Table 15, levels of borrowing from family/friends are lower in the present study than those found by the Global Findex, and slightly lower than those found in other Pacific island studies. This may be because the financial capability study (and the financial competency studies in several Pacific island countries) has only examined borrowing from family and friends that must be repaid and does not include loans which create a reciprocal, rather than financial, obligation. Store credit is an important form of short term finance, in particular for rural households. Anecdotal evidence indicates store credit is typically interest free and the duration is typically days or, at most, weeks. The characteristics of stores which provide store credit (or do not provide store credit) and importantly, the impact of store credit on the viability of trade-stores is not well understood. The levels of store credit found by the present study are significantly lower than those found in other studies in PNG and other Pacific island countries. To some extent this may be due to location differences; the Morobe/Madang study is the only study to explore financial Table 15  Informal Sector Borrowing Morobe/ Madang PACIFIC Informal loan (money lender)  6% 6–9% Loan from family/friends (that has to be repaid)  9% 12–14% Store credit 15% 32–36% 51 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea behaviour in remote rural communities. To some extent it may be due to sampling (the PFIP financial competence studies only examined low income communities). To some extent the lower level of reported use of store credit may be due to a (common) misunderstanding that purchase of goods with deferred payment is a form of credit. Overall 34 percent of urban households reported formal borrowing and 24 percent informal borrowing. By contrast 3 percent of rural households reported formal borrowing and 18 percent informal borrowing. It would seem there has been little change in rural households’ borrowing behaviour over the past twenty to twenty five 39 years. Fernando, in a review of informal finance in PNG (1992), noted that in rural areas the major sources of credit appeared to be members of borrowers’ extended families and clans, and trade-store owners who extend credit for consumption purposes. 4.5. Responsibility for Selection of Financial Products Overall, 37 percent of household financial decision makers stated they were involved in the selection of financial products used by the household. However, thirty nine percent (39 percent) of households stated no one was responsible for financial product selection. There were significant gender differences within households. Approximately half of the men who made financial decisions on behalf of their household stated they were involved (individually or jointly with their partner) in the selection of household financial products. By contrast only a quarter of women stated they were involved in financial product selection. A further 28 percent of women stated they were not involved in the selection of financial products and that this was an activity for which their partner was responsible (less than 2 percent of men stated their partner was solely responsible). The gender difference was further amplified in urban households. Sixty eight percent (68 percent) of men living in urban households stated they were solely responsible for the selection of the household’s financial products, compared to 14 percent of women. Overall women are less involved in all forms of household financial product selection. Women’s principal source of livelihood appears to play a role in encouraging active engagement in household financial product selection (as it does in with participation in the formal financial sector generally). As shown in Table 16, whilst overall women were more likely to state their partner was responsible for financial product selection, women employed in the formal sector were three times more likely to state they were 39  Cited in Conroy (Conroy, 2000, p. 230). 52 Financial Inclusion in Morobe and Madang Table 16 Responsibility for financial Product Selection Male Female Informal Formal Informal Formal sector sector sector sector Respondent only 30.7% 63.0% 5.6% 49.3% Respondent (solely or with 43.6% 87.5% 17.9% 58.1% someone else) Respondent’s partner (solely or 2.0% 4.5% 29.8% 27.2% with someone else) Nobody at all 46.8% 8.0% 45.9% 14.7% involved in the selection of household financial products than women working in the informal sector. Overall, formal sector employment also appears to influence active engagement in the selection of financial products. Formal sector employees were 3–6 times less likely to state no one in the household was responsible for financial product selection than those who derived their income from the informal sector or self-employment. 53 5.  Financial Capability in Morobe and Madang Financial capability encompasses a broad range of financial knowledge and skill, behaviour and attitudes. As discussed in Chapter Two, the present study has sought to develop an understanding of the financial capability of adults in Morobe and Madang. The Morobe and Madang dataset will contribute to the national dataset. The following analysis has examined financial capability, with a particular focus on gender differences, differences between rural and urban households and differences between respondents whose primary source of income was from formal sector employment and respondents whose primary source of income was from the informal sector or self-employment. Four aspects of financial capability have been examined: the planning and management of current cash flows, the planning and management of future cash flows, responsibility for management of the household’s cash flows, and interactions with financial institutions. 5.1. Managing Current Household Cash Flows Planning and Budgeting Household Cash Flows Household Plans and Budgets Most households (60 percent) reported planning how income would be used. However, most households (55–75 percent) that reported planning the use of income also reported the plan was only a ‘rough plan’. In addition, most households (70 percent) only planned sometimes and less than a quarter of households stated they always kept to the plan. Households in urban communities and those receiving formal sector wages or salary were more likely to report planning cash flows (75 percent of households) than rural households and households receiving informal sector income as the primary source of income (55 percent of households), and were more likely to state that management of cash flow was always planned (50 percent of households). Overall, whilst many households considered they planned cash flows; in effect households that planned typically did so informally and intermittently. A similar picture emerges in respect to household budgeting. Between 40 percent and 70 percent of 54 Financial Capability in Morobe and Madang households reported the household had a budget to manage household cash flows. Households living in urban communities and engaged in the formal sector employment were more likely to budget. However, most budgets were not written down. In general, it appears most households in Morobe and Madang endeavour to plan and budget household cash flows. Planning and budgeting is therefore an activity which is acknowledged as important. However most plans and budgets appear to be general and are neither written down nor usually adhered to. Planning Horizon There was a difference in the planning horizon between respondents whose primary source of income was formal sector employment and respondents whose primary source of income was informal or self-employment. As shown in Figure 22 approximately 25 percent of respondents stated they planned ahead six months (or more) and 12–14 percent of respondents stated they planned several months ahead. The planning horizon appears to be influenced by the income cycle. Respondents who derived their income principally from formal sector employment were more likely to adopt a planning horizon spanning a week or several weeks. This is likely to coincide with the wage or salary cycle. By contrast respondents who derived their principal income from the informal sector were more likely to state they did not plan ahead. Figure 22  Planning Horizon Does Not Plan 30.0% 25.0% 20.0% 15.0% Six Months or More 10.0% Days 5.0% 0.0% Months Weeks Informal Sector/Self-Employed Formal Sector 55 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Responsibility for Management of the Household Budget In most households at least one person accepted responsibility for management of the household budget. Less than 2 percent of households stated no one was responsible for the management of the household budget. Men appear to be more engaged in planning and budgeting household cash flows than women and were more likely to state they were responsible for the management of the household budget than women. Overall 88 percent of men, compared to 57 percent of women stated they were responsible for the household budget, either solely responsible or with their partner. Women were 2.5–3.0 times more likely than men to say they did not plan ahead. The lower level of engagement by women in the management of household finances is consistent across all aspects of financial capability examined by the financial capability study. It is not simply that women are less likely to own a bank account than men, but that, overall, women are less engaged in the management of money than men. Confidence with the use of English in communication and primary source of employment may contribute to a lower level of formal financial product ownership. However, irrespective of, for example, primary source of income, women appear to have a subordinate role in the management of household finances. Management of Household Expenditure Management of Cash Surplus Slightly more than half of the households stated they had money left over after meeting household expenses at least some of the time. Households with formal sector income were more likely to state the household has a surplus than households which did not earn formal sector wages or salary (87 percent of households earning formal sector income, compared to 64 percent of households with only informal sector income). The pattern was similar for urban and rural households, and for men and women. Households typically saved the surplus to provide for unforeseen expenses for example emergencies or medical fees (30–50 percent of households), or for food and other necessary items (30–45 percent of households). Households also spent the surplus on food and other necessary items (20–35 percent of households. Interestingly, 20 percent of households earning formal sector income stated the surplus was used to support other family members. Management of Cash Deficit By contrast 80–90 percent of households reported running short of money after meeting household expenses. Households which only had informal sector income were likely to state the house ran short of funds. As shown in Table 17, a range of causes were cited. All households cited a range of expenditure related causes. Households whose principal source of income was informal sector or self-employed 56 Financial Capability in Morobe and Madang Table 17 Reasons for Household Cash Shortage Urban Township Rural Informal Informal Informal Sector/Self- Formal Sector/Self- Formal Sector/Self- Formal employed Sector employed Sector employed Sector Insufficient/low income 33% 18% 31% 61% 42% 23% Fluctuating/unreliable 6% 0% 27% 4% 35% 18% income Unexpected expenses/ 38% 22% 23% 22% 11% 32% events Increased cost of food 20% 26% 19% 14% 18% 27% and other necessary items Need to provide financial 24% 23% 15% 14% 6% 37% help to others Overspending 31% 44% 27% 37% 16% 34% Failure to plan ahead/ 31% 26% 31% 11% 34% 24% budget also, not surprisingly, cited income related causes. However, only 4 percent of rural households cited the inability to get to the market as a principal cause of cash shortage. Failure to plan was one of the most commonly cited causes of the household running short of money. The most common action taken when the household ran short of funds was to borrow money from family or friends, ‘sell something’ (particularly rural households) or to simply go without essentials. Between 17–30 percent of households earning formal sector income also stated they would borrow from a money lender, or from their employer. Overall, there appears to be a different pattern used by urban and rural households in respect to funding cash shortfalls. All households stated they would seek to borrow from family or friends. Urban households tended to also state they would borrow money, whereas rural households tended to state they would seek to earn more money. Overall, however the use of credit to pay for food or other necessary items because the household had run short of money was common. Between 50–60 percent of households, irrespective of location or primary source of income, reported using credit or borrowing money to buy food if there was a cash shortfall. The pattern of urban households being more willing to borrow money to manage cash shortfalls is also reflected in the use by urban households of credit to repay existing debts. Thirty percent (30 percent) of urban households stated they borrowed money to 57 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 18  Household Borrowing Limit Urban households, Rural households, formal sector informal income income I/we could afford to borrow more if  9.4% 30.8% I/we wanted or needed to I/we have borrowed to my/our limit 46.0% 44.6% and could not afford to borrow more I/we have borrowed more than I/we 39.4% 22.2% can really afford repay outstanding obligations and were 2–3 times more likely to state money would be borrowed than rural households. Men were more likely to state they borrowed to pay existing debts, reflective perhaps of the greater control exercised by men over household cash flows and financial obligations There is a perceptual difference in the ability to borrow additional funds. As shown in Table 18, urban households earning formal sector wages or salary tended to consider they or their household could borrow more money if required and were less likely than rural households to consider the household had borrowed more money than could be afforded. Overall, women appear to be less confident about household borrowing than men. Sixty percent (60 percent) of women considered they or their household had borrowed to the limit, compared to 25–35 percent of men. Knowledge of Cash Position and Spending As shown in Figure 23, there is a difference between urban and township households (which have a higher proportion of adults earning formal sector income) and rural households (which earn primarily informal income). In addition to being more likely to plan ahead, urban households appear to be more likely to know the household’s cash position and how much has been spent, both individually and in respect to the household generally. This is indicative of a pattern: households in urban or township communities are more focused on money, and within this context the monitoring and management of money, than rural households. Self-Discipline Cash Flow Management Respondents were asked how disciplined they considered they were at managing money. This is not a measure of self-discipline per se, but an indication of respondent’s confidence in their ability to control their management of money. Three aspects of self-discipline were considered: how disciplined the respondent was at managing 58 Financial Capability in Morobe and Madang Figure 23  Knowledge of Spending 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Does Not Know How Much Does Not Know How Much Does Not Know How Much Money Was Spent Personally Money the Household Money You and Your in the Last Week? Spent Last Week? Household Have Available for Day-to-Day Spending? Urban Township Rural money generally, whether the respondent learned from other’s mistakes at managing their money, and the respondent’s propensity to spend on unnecessary items. The pattern of responses was similar: most adults considered they were very disciplined at managing money; that they learned from other’s mistakes and only rarely, or at most occasionally, that they bought things they knew were not necessary before buying essentials, or bought unnecessary items even when they knew they could not afford the item. Urban respondents tended to regard themselves as more disciplined than rural respondents. Men tended to regard themselves as more disciplined than women did, and less likely to purchase unnecessary items. 5.2. Planning Future Household Cash flows Three aspects of the planning future household cash flows were examined: planning for major future expenses (both expected and unexpected), planning for the children’s future and planning for older age. Planning for Major Future Expenditure Expected Future Expenses About half of the households surveyed stated they expected to incur a major expense (equivalent to at least one month’s household income) at some time over the coming year. Households in urban communities, in particular households earning formal sector wages or salary as a primary source of income were more likely than rural households 59 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea to state the expense could be met without borrowing (50–60 percent of urban households compared to 40 percent of rural households). Households with a savings account were slightly more likely to state the expense could be met without recourse to borrowing. Most households which stated the expense could not be met without borrowing also stated they had not done anything to be able to meet the expense (50–60 percent of households). There was little difference in responses between men and women, although women were more likely to state they were worried about meeting the expense. Rural households were also more likely to express concern about their ability to meet the expense. Unexpected Future Expenses Overall, 70 percent of households stated they would not be able to cover a major unexpected expense without borrowing. Most households (77 percent) had not done anything to be able to meet an unexpected major expense. Whilst ownership of a savings account did not appear to be a significant factor in household’s ability to meet expected expenses, households with a savings account were more likely to state the household could pay a major unexpected expense (equivalent to at least one month’s household income) without having to borrow money. Seventeen percent (17 percent) of households that did not own a savings account stated they expected to be able to cover the expense, compared to 35 percent of households with a savings account. Nearly all households (90 percent) expressed concern about the household’s ability to cover a major unexpected expense. Planning for Children’s Future Households typically had been between 3–3.5 dependent children. Rural households tended to have slightly larger families than urban households. Respondents were 40 asked if the household had planned, or was planning for the children’s future. Most households reported planning for their children’s future in some way. Urban households were more likely to report they were planning than rural households (90 percent of urban households compared to 70 percent of rural households). Households which were planning for their children’s future typically used more than one form of provision. Table 20 shows the percentage of households which reported using each strategy (‘cited‘) and for each strategy the percentage of respondents who also reported using other means of provision for the children’s future. As shown in 40  Responses were pre-coded and read to respondents. 60 Financial Capability in Morobe and Madang Table 19  Planning for Children’s Future Also Cited Cited 1 2 3 4 5 6 1. Saving money for children’s education 44% 37% 15% 13%  9% 41% 2. Saving money to pass on to children 20% 84% 26% 20% 15% 41% 3. Investing money to pass on to children  8% 82% 62% 23% 29% 38% 4. Investing in land/buildings to pass on to  9% 64% 42% 21% 25% 43% children 5. Investing in business to pass on to  6% 67% 49% 41% 39% 30% children 6. Planning for children’s future in other 33% 56% 24% 10% 12%  5% way Table 19 the most common form of planning for the children’s future was to save, in particular to save for the children’s education. One third of households stated they were planning for the children’s future in an unspecified way. However, over half of these households also reported saving for education. The second most common strategy was also savings based (saving money to pass on to the children). As shown in Table 20, most households used cash based provision for the children’s future. Similar to the longer planning horizon adopted for the management of household cash flows by households with formal sector wages or salary as a principal source of income, these households were more likely to cite longer term strategies (investment) than households which relied on informal sector income or self- employment. Table 20  Planning for the Children’s Future by Principal Source of Income Informal Formal Sector/ Sector Self-employed Saving money for children’s education 56% 40% Saving money to pass on to children 31% 17% Investing money to pass on to children 30%  4% Investing in land/buildings to pass on to children  8%  9% Investing in business to pass on to children 10%  4% Planning for children’s future in other way 28% 33% 61 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Planning for Older Age The generic financial competency survey includes a section examining the strategies used to plan for older age when the respondent is no longer working due to age. The section has two paths: one for respondents under the age of sixty, the second for respondents over the age of sixty. The difference in the paths being that the questions for respondents over the age of sixty assume the respondent is no longer working. It was acknowledged during the development of the instrument that the use of a common age for ‘retirement’ was somewhat arbitrary. This approach used by the generic financial capability survey to developing an understanding of planning for older age is problematic for PNG for two reasons. Firstly, 41 life expectancy in PNG is 60 years men and 65 years women, with the probability of dying between 15–60 years of 31.9 percent for men and 24.3 percent for women. In absolute terms, therefore, the number of Papua New Guineans over the age the sixty is likely to be very low. This has been the case with the survey in Morobe and Madang provinces. Only 5.4 percent of respondents were 60 years or older. Secondly, and perhaps more importantly within the context of many Pacific island countries, the concept of retirement is derived primarily from the emergence of formal employment and social security in Europe. Retirement provision has become an important policy issue in many developed countries, in particular countries with an aging population and in which most adults earn salary or wage income and the state typically makes some form of provision for citizen’s old age. However, the concept of ‘retirement’ from the workforce does not have the same meaning in the context of Morobe and Madang in which most people live in a rural community which has a high subsistence and self-generated income component and have no expectation of becoming eligible for a pension, whether state or private provision, at a specified age. There is currently only one old age social pension scheme in PNG, the pension scheme provided by New Ireland province for residents sixty years and older. The provident funds in PNG are solely for wage and salary earners. Typically in Papua New Guinea, people must work until they are no longer able to work. The generic English language question used by the survey is; ”what strategies do you have for meeting your/your household’s expenses in your old age?” The issue is not age per se, but how people expect to meet expenses when they are no longer able to work due to age. The question used for no longer working used the following To Pisin construct: “taim yu no wok moa bikos yu lapin pinis,” with ‘old’ being self-determined. Overall, a significant percentage of respondents stated they had no strategies to meet their or their household’s expenses when they were no longer able to work due to 41  www.who.int/countries/png/en/. 62 Financial Capability in Morobe and Madang age. Forty four percent (44 percent) of respondents had no strategy and therefore did not know how they would meet expenses when they were no longer able to work. Table 21 shows respondents’ expected means of meeting expenses when they can no longer work due to age. The table shows the percentage of respondents who reported each expected strategy (‘cited‘) and for each strategy cited the percentage of respondents who reported they also expected to use other strategies to meet expenses. As shown in Table 22 most respondents expect to use multiple sources to fund their expenses. The issues associated with the concept of ‘not working’ can be seen in the pattern of responses. The most common response was that respondents expected to earn income from a business. As shown in Table 23 this response was frequently provided by men who derived their principal source of income from the informal sector or self-employment and who already earned business income. Approximately one quarter of respondents expected to receive support from others. The combination of financial support and business income were the most common strategies respondents expected to use to fund expenses when they were no longer working. Most respondents stated the strategies they expected to use to fund expenses when they were no longer working were already in place. As shown in Table 22, there were important differences in the pattern of responses. Respondents whose principal source of income was informal sector were significantly less likely to state they would use savings or long term savings (for example provident fund) to meet expenses. Men were significantly more likely to state they would use an inheritance to meet expenses. The reasons why only men stated they would be able to access inheritance Table 21  expected MEANS to meet expenses when no longer working due to age Also Cited Cited 1 2 3 4 5 6 7 8 9 1. Financial help/support 26% 12% 6% 2% 0% 10% 31% 36% 2% 2. Savings/other financial assets 13% 25% 17% 3% 6% 12% 18% 48% 7% 3. Employer superannuation 7% 25% 33% 6% 6% 0% 17% 29% 10% 4. Other superannuation 2% 33% 25% 25% 8% 0% 34% 27% 0% 5. Insurance 1% 0% 100% 60% 20% 20% 0% 19% 20% 6. Sale of non-financial assets 8% 34% 20% 0% 0% 2% 29% 58% 8% 7. Inheritance 13% 61% 17% 9% 4% 0% 17% 48% 5% 8. Business 31% 31% 20% 6% 1% 0% 15% 21% 7% 9. Will always work 3% 20% 35% 26% 0% 6% 24% 24% 79% 63 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 22  Current and Expected Strategies Male Female Informal Sector/ Informal Sector/ Self-employed Formal Sector Self-employed Formal Sector Expect Expect Expect Expect to have Has now to have Has now to have Has now to have Has now 1. Financial help/ 25% 22% 21% 20% 27% 24% 20% 20% support 2. Savings/other 10% 10% 40% 36% 4% 3% 20% 24% financial assets 3. Employer 2% 2% 37% 38% 2% 2% 23% 26% provident/ superannuation 4. Other provident/ 1% 0% 11% 12% 0% 0% 6% 6% superannuation 5. Insurance 0% 0% 6% 5% 0% 0% 0% 0% 6. Sale of non- 11% 11% 2% 4% 4% 3% 0% 0% financial assets 7. Inheritance 20% 18% 16% 15% 9% 7% 0% 0% 8. Business 41% 39% 28% 23% 20% 19% 15% 9% were not explored by the survey. It is possible this is due to societies in Morobe and Madang being primarily patriarchal (which may also explain the higher level of business income cited by men). As shown in Figure 24, respondents who earned formal sector wages or salary were more likely to consider the strategies they expected to use to fund themselves and their household when they are no longer working would provide for all expenses. Less than 50 percent of those earning informal sector incomes expected to be able to meet expenses. Women expressed particular concern. Less than a third of women considered they would be able to meet expenses when they were no longer working. Overall respondents earning formal sector income were significantly less worried about meeting expenses. Twenty four percent (24 percent) stated they were very worried about meeting expenses, compared to 45 percent of respondents earning informal sector income or self-employment. Approximately 4.6 percent of the Papua New Guinea population is aged 60 years and older (World Bank, 2013c). Assuming a similar percentage of the working age population has no strategy to fund their expenses when they are no longer working as has been found for Morobe and Madang, it is likely that approximately 1.9m adult Papua New Guineans do not know how they will meet their expenses in old age. It is 64 Financial Capability in Morobe and Madang Figure 24 Expectation Strategies Will Cover All Expenses 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Male Female Male Female Formal Sector Informal Sector/Self-Employed Will Provide Enough Money to Cover Expenses Will Not Provide Enough Money to Cover Expenses Has No Strategies estimated the population aged 60+ in Melanesian countries will rise to 7.2 percent by 2025 (Hayes, 2009). Assuming a population growth rate for PNG of 1.89 percent p.a., this suggests that this figure could rise to approximately 2.3m Papua New Guineans by 2025. Funding expenses in retirement appears to be an emerging policy issue for PNG, in particular for adults living in rural areas. The frequently made assumption that family or clan groups will provide for the elderly may no longer be appropriate. 5.3. Responsibility for Household Financial Management Adults in Morobe and Madang who are responsible for making financial decisions on behalf of their households are more likely to state they are responsible (either solely or with someone else in the household—usually their partner) for the management of shorter term rather than longer term or less certain expenditure. As shown in Figure 25, the further out the expenditure horizon, or the less certain the expenditure, the greater the likelihood the respondent will state that either their partner is responsible for the expenditure, or that no one is responsible. Men are significantly more likely to state they are responsible for the management of all aspects of household finances than women (refer to Figure 26). There is no major difference in responses relative to the respondent’s confidence at communicating in English, to the respondent’s principal source of income, or whether the respondent lived in a rural or urban household: men dominate the management of household finances. Women living in rural households are even less likely to state they are responsible for the management of household finances than women living in urban 65 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 25 Responsibility for Management of Household Expenditure 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Setting and Managing Your Managing Regular Managing the Managing Managing the Managing the Household’s Household Household’s One- Requests for Household’s Household Budget Spending on Expenses Off Expenses Financial Financial Essential Items Assistance Documents Respondent Partner/Someone Else No One households. Only one aspect of household financial management was consistent across gender: men and women give similar responses in respect to whether no one was responsible for an aspect of household financial management. In other studies in Pacific island countries a different pattern has been found. Men and women were likely to share responsibility for day-to-day expenditure, with men being more likely to have greater responsibility for less frequent expenditure. Figure 26 Responsibility for the Management of Household Finances by Gender 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Setting and Managing Your Managing Regular Managing the Managing Managing the Managing the Household’s Household Household’s One- Requests Household’s Household Budget Spending on Expenses Off Expenses for Financial Financial Essential Items Assistance Documents Male Female 66 Financial Capability in Morobe and Madang 5.4. Financial Knowledge The financial knowledge required to effectively use formal financial services and to manage cash flows is a complex topic. The scope and complexity of financial knowledge required by financial services consumers is related to the level of engagement with the formal financial system. Evidence from the present study indicates there a wide range of levels of engagement with the formal financial system, ranging from very limited engagement by many households in rural areas, particularly rural communities with limited access to finance points, to multi-faceted engagement by households whose principal source of income is formal sector salary and who regularly use electronic transaction services, savings, long term savings (including provident or superannuation), and formal credit (including, importantly, secured credit). Two fundamental aspects of financial knowledge required by financial services consumers irrespective of their level of engagement with the formal financial sector have been examined: the process used to select financial services products and services and understanding of the cost of financial services. Selection of Financial Products In respect to the selection of financial services products and services, several aspects of the selection process are critical: whether consumers seek to find a product which best met their needs, whether information is sought about the product from a range of sources not just the financial services product provider, whether alternative products were considered and whether consumers checked the terms and conditions prior to committing to purchase the product. The range of financial products and services available to urban customers is more extensive than that available to rural customers. In addition, the ability to be able to seek information and to consider terms and conditions is influenced by functional English literacy. As shown in Figure 27, urban consumers who are confident in their ability to communicate in English appear to be more likely to search for information from a range of sources, to consider product alternatives in order to find the product that best meets their needs and to check terms and conditions before purchasing the product. Consumers who are not confident in their ability to communicate in English are less likely to be critical in their product selection and are very likely to purchase a product without reviewing the terms and conditions, irrespective of whether they live in an urban or rural community. Limited ability to communicate in English creates vulnerability in product selection. The present study has not examined consumer protection as this is still developing in PNG. However it is evident there are significant levels of potential consumer 67 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Figure 27 Selection of Financial Products 80.0% 70.0% 60.0% 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% Searched for Considered Searched to Find Did Not Check Product Information from a Alternatives Product Best Met Terms and Conditions Range of Sources Needs Urban Communicate in English Rural Communicate in English Urban Cannot Communicate in English Rural Cannot Communicate in English vulnerability for consumers who have a limited ability to read terms and conditions written in English and/or who live in rural communities with very limited product choice. Understanding the Cost of Financial Services Understanding the cost of money is a fundamental financial capability, irrespective of the consumer’s level of engagement with financial institutions (whether formal institutions or informal institutions). As shown in Table 23, most adults in Morobe and Madang do not know how much interest has been paid on borrowing, how much interest has been earned on savings and, in respect to formal financial institutions, the fees paid for financial services. This also suggests there may be considerable vulnerability in the use of financial services by adults in Morobe and Madang and, related to this, significant risk of exploitation. Table 23  Knowledge of the Cost of Financial Services Do not know how much interest has been paid on loans over the 82% past year Do not know how much interest has been received on savings over 86% the past year Do not know the fees you have paid on financial products over the 82% past year 68 Financial Capability in Morobe and Madang 5.5. Managing Relationships with Financial Institutions Most people (66 percent) stated they never went anywhere when they needed to seek financial advice. Those who had sought financial advice reported seeking advice from a broad range of sources. Approximately a quarter of those who had sought advice reported seeking advice from a bank or other financial services professional. Other sources of advice included family, friends or spouse. Few people (10 percent) reported having had a dispute with a financial services provider. Most people who had had a dispute did not endeavour to resolve the issue with the financial institution providing the product. They reported either doing nothing or talking to family or friends. The potential vulnerability of financial services consumers in Morobe and Madang is further underscored by the limited understanding of the risks associated with the use of financial services providers. Eighty one percent (81 percent) of adults stated they had never tried to find out about the risks of using the financial services providers they currently use or previously used. Overall most financial services consumers do not appear to search for financial services, check product terms and conditions, understand the cost of the financial services they are using or the risks of the financial services providers whose products they use and do not seek financial advice. 5.6. Findings Relative to Other Financial Capability Studies The financial capability study in Morobe and Madang is the only financial capability population study undertaken to date in a Pacific island country. There have been several other studies which had had a more limited scope and have focused on poor and low income households. The studies were undertaken by central banks in Samoa, Fiji, the Solomon Islands and PNG, with the support of the Pacific Financial Inclusion Programme (UN-PFIP). The focus of the Pacific island studies was similar to the present study in that the studies sought to understand levels of financial inclusion, financial knowledge and skill and financial behaviour. The methodology was slightly different. The studies sought firstly to understand the financial competencies adults in low income households, who made financial decisions on behalf of their households, considered they needed to have in order to manage their household’s finances effectively, and to then measure the extent to which adults were able to demonstrate competence. A common competency set (Sibley & Liew, 2011) was used for the studies, as similar 69 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea competencies were considered to be necessary by low income households in each country. The PNG Financial competency study examined the financial competence of low income households in several settlements in Port Moresby, villages in Mekeo and Galley Reach. The findings from the studies are broadly similar across countries and with the findings from the present study: ■■ Most low income households owned few financial products. Financial products owned were principally savings products. There was very limited use of formal credit. ■■ Rural households were significantly less likely to own formal financial products and services than urban households and exhibited lower levels of financial capability/financial competence. The findings of the PNG financial competence study in respect differences in the level of bank account ownership were similar to the findings from the present study. ■■ In Fiji (in respect to both iTaukei and Indo-Fijian households), the Solomon Islands and PNG, women were consistently less likely to be involved in the management of household finances (other than the management of day-day expenditure) and were less likely to own formal financial services. ■■ Many adults did not know how they would meet expenses when they were no longer working—other than relying on their children or family or friends. ■■ The overall level of financial knowledge was low and most adults had very limited understanding of the cost of financial services. There was a significant difference in the use of informal store credit reported by the financial competence studies compared to the present study. Use of store credit, in particular the use of store credit by rural households, was typically at least twice the level reported for Morobe and Madang (32–36 percent compared to 15 percent for Morobe and Madang). The reasons for the difference level in the use of store credit may be, as previously discussed, due to lack of understanding that deferred purchase of goods is a form of credit. It is also possible that the difference in the respective geographic and population scope of the studies may result in lower levels of store credit being found by the present study. The Morobe Madang study includes a significantly broader range of households, in particular households earning formal sector income, than the PFIP studies and also has a wider geographic spread of sampling locations. It is also possible that store credit is not prevalent in Morobe and Madang and that a greater use of store credit will be found in other locations as the national financial capability study progresses. 70 6.  Implications for Policy and Strategy The findings from the study of financial inclusion and financial capability in Morobe and Madang are discussed in relation to the Maya Declaration Goals which form the basis for financial inclusion and financial literacy policy in PNG and provide the framework for developing strategies and implementing programmes to increase financial inclusion and financial capability in PNG. This chapter does not make recommendations in respect to possible changes to the National Financial Inclusion and Financial Literacy Strategy. The scope of the present study is limited to Morobe and Madang and it is not appropriate to suggest change or development to a national strategy based on a survey of two provinces. However, on the basis the findings from the present study are likely to be indicative of the situation in other provinces; potential implications for the achievement of the Maya Declaration Goals are discussed. Goal: To reach one million more unbanked low-income people in Papua New Guinea, 50 percent of whom will be women Situation in Morobe and Madang The study of financial inclusion in Morobe and Madang has found a significant difference between urban and rural communities. This is consistent with the findings from the financial competency study in Central province. Evidence from Morobe and Madang suggests levels of financial inclusion in urban communities may be approaching, or may be at levels found in other developing countries. The study also suggests urban households, in particular households in which the principal source of income is formal sector wages or salary, are engaging with the formal financial system across a broad range of product groups, extending beyond savings/transaction accounts to long term savings and formal credit. By contrast levels of financial inclusion in rural communities, across all product groups, are very low. The study has also found a significant difference between levels of financial inclusion for men and women, particularly in urban communities. Access to financial services appears to have resulted in increased levels of financial inclusion primarily for men. Women in urban communities are significantly more likely to have a bank account than women in rural communities and may be accessing financial services indirectly 71 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea by accessing a bank account owned or controlled by a male. However, relative levels of financial exclusion by women living in urban communities are significantly higher than those found in rural communities. Implications for Policy and Practice Households need access to financial services infrastructure. Most households in PNG are rural. The findings from the present study suggest that many rural communities, effectively, have no or very limited access to formal financial services. It is possible the Maya Declaration Goal may be achievable in urban communities. However, the findings from Morobe and Madang suggest that, without major development to the rural financial services infrastructure, bringing large numbers of currently financially excluded Papua New Guineans into the formal financial system will not be possible. Increasing access to financial services does not appear to benefit women to the same extent as men. If the finding from the study in Morobe and Madang is found to be the case across PNG, it is likely that, unless there is a substantive change to financial inclusion programmes, and within this context possibly also products and services (whether public or private sector), the goal of gender equity will probably not be achieved and the gender gap may even be further exacerbated. Further investigation is required to understand why women are less likely to own any form of formal financial service, even when the financial service is accessible and likely to be affordable. Goal: BPNG to lead efforts to create a financially competent generation of Papua New Guineans through financial education and financial literacy Situation in Morobe and Madang Most households in Morobe and Madang appear to endeavour to plan and budget cash flows. However, for most households, irrespective of whether the household receives regular formal sector wages or salary, or more unpredictable informal sector or self-employed income, plans and budgets are typically informal, focused only on major items and are not documented. Whilst most households may consider they are planning cash flows, in effect the household has priorities for spending. Overall, households in urban communities, in particular households which receive regular wages or salary, are more likely to pro-actively manage household finances. This is not surprising as these households engage with the money economy on a daily basis, across a broad range of financial activities. Money is embedded in the life of urban households to a greater extent than in many rural households. It tempting to regard the lower level of money management by rural households as evidence rural households have lower levels of financial capability. This may not, however, necessarily 72 Implications for Policy and Strategy be the case. Capability is relative to financial activity and it is possible rural households do not need the same level of financial knowledge and skill as is required by urban households. Irrespective of location or source of income, women are less involved in all aspects of household financial management measured by the study than men and have less responsibility for the aspects management of household finances measured by the study than men. Whilst many respondents stated household finances were managed jointly, in effect the management of household finances in Morobe and Madang appears to be dominated by men. The financial competence study in Central province found a similar pattern of responsibility. Many adults, in particular rural households and women, do not know how they will fund their and their household’s expenses when they are no longer working. Access to provident schemes and formal financial services promotes savings. However, for most adults in Morobe and Madang (and potentially PNG generally), there is little opportunity to make progressive provision for old age. Implications for Policy and Practice As the study of financial capability in PNG progresses, it may be appropriate to consider the development of rural and urban capability indices in addition to a national index. Given the likelihood that budgeting and planning household cash flows is not common across households in PNG, the focus on budgeting and planning which is a common feature of financial literacy programmes would appear to be appropriate. As the use of formal credit expands (and evidence from the present study indicates that formal sector employment may facilitate the use of formal sector credit), inclusion of responsible borrowing modules in financial literacy programmes may also be warranted. It is also suggested that consideration may need to be given to including longer term savings or asset accumulation modules in order to increase awareness of the need to prepare to older age and how gradual longer term savings can be achieved. Findings from the present study and the financial competence study suggest there may be a requirement for financial literacy programmes specifically for women. Increasing formal employment and monetisation will inevitably result in an increasing need for Papua New Guineans to provide for their old age. If the findings from the study in Morobe and Madang are replicated across PNG, it is possible that the scope of financial inclusion programmes may need to expand to include longer term planning and the preparation for old age when it is no longer possible to work. In a wider context, the ability to meet expenses in old age is an emergent policy issue in 73 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea PNG. Findings from the present study and the national financial capability study, once completed, may provide a useful input to the policy discussion. Goal: To actively support innovative use of technology for scaling-up access to financial services and financial literacy Situation in Morobe and Madang Branch, ATM and EFTPOS networks in PNG are largely urban. This is the case for Morobe and Madang. Agent networks do not, at this time, appear to be sufficiently dense to enable most rural households to readily access formal financial services. Mobile phone telephony appears to be an important component of extending financial services, in particular to rural communities. However, in Morobe and Madang, there appears to be a significant difference between rural and urban communities in respect to adults’ capacity to use the mobile phone, and by extension the potential capacity to use the mobile phone for financial services. Levels of mobile phone ownership, combined with adults’ ability to use the mobile phone to text as well as to make calls, suggest mobile phone banking has significant short term potential in urban communities, and rural communities with access to cell-phone telephony. Implications for Policy and Practice Developing extensive branch networks or even extensive ATM networks in rural areas is not an option for PNG. Other than BSP Rural, commercial financial services providers are not expanding services into rural areas but are focusing on urban communities. It may be appropriate to consider the development of a financial services delivery architecture for rural communities in PNG. Alternative delivery channels are likely to be required. There appears to be broad acceptance that the mobile phone has a key role in expanding financial inclusion in PNG and BPNG has been pro-active in facilitating a supportive regulatory environment. Whilst there is some evidence of a correlation between access to formal financial services and financial inclusion, the findings from the present study suggest that an approach based on ‘build it and they will come’ may not be appropriate for mobile phone based financial services. The combination of more limited ownership or access to a mobile phone and mobile phone capability in rural communities, in particular for women, suggests the expansion of mobile phone based financial services (as opposed to the use of mobile telephony to deliver agent based financial services) in rural areas may require concurrent capacity building Goal: To strengthen consumer protection by issuing prudential guidelines and creating a platform for various national regulators and industry networks to monitor consumer protection 74 Implications for Policy and Strategy Situation in Morobe and Madang Financial services consumers in Morobe and Madang appear to have a limited understanding of the cost of the financial services they use. Many consumers, in particular consumers who are not confident in communicating in English, do not search for information about financial services before committing to buy a product, they do not consider alternatives or look at alternative products and, perhaps most importantly, they frequently do not check terms and conditions of the products they purchase. The financial competence study in Central province found a similar low level of critical purchase behaviour and limited understanding of the cost of formal financial services used. Implications for Policy and Practice Findings from the present study, and the earlier study in Central province, suggest consumer awareness is limited for many adults and there is therefore potential vulnerability to predatory practices. This not only exposes households to the risk of high interest and fee charges, but also creates a potentially significant vulnerability to financial scams. Whilst few households appear to have had a dispute with a financial services institution, findings from the present study suggest a platform to monitor consumer protection is likely to be warranted. Goal: To begin the process of integrating financial inclusion in local and national government, including getting the National Executive Council to endorse the National Financial Inclusion and Financial Literacy Strategy by quarter 4 of 2013 The focus of the present study is not relevant for this goal. Goal: To promote regular collection and use of financial access data to inform policy making and help identify key dimensions of financial inclusion in Papua New Guinea Situation in Morobe and Madang BPNG has been implementing a programme to collect enhanced supply-side data. Collection of field data, in particular field work in rural communities, can pose significant challenges. The reasons for the challenges are well known. Nevertheless, there is a very real need to better understand financial exclusion in PNG and financial capability. Despite initial challenges, the present study has been successful in collecting field data, including field data from remote communities, at an acceptable cost. Implications for Policy and Practice Findings from the present study have contributed to the understanding of financial inclusion and financial capability in PNG. The findings are not, however, national. It 75 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea is recommended the national financial capability study be completed with periodic follow up to determine change over time. A national baseline is required. Goal: To optimize these results through knowledge sharing and effective coordination of stakeholders, including development partners, by the newly established Centre of Excellence for Financial Inclusion chaired by the Bank of Papua New Guinea The focus of the present study is not relevant for this goal, other than the dissemination of findings. 76 Appendices Appendix 1: Overview of Districts Morobe Province Morobe province is divided into nine (9) districts, including Lae (refer to Map 4). The survey was conducted in six (6) of these districts, namely Menyamya, Huon Gulf, Bulolo, Finschafen, Markham, and Lae. Lae Lae city had a reported population of 148,934 in the 2011 Census, and is, as stated, the hub of much of the Momase and Highlands region. It is the largest port in the South Pacific, by far, outside Australia and New Zealand, with a major recent upgrade, notably with ADB financing support. As well as serving as export port for the Highlands and Markham/Ramu agricultural industry, and the mines of Porgera, Hidden Valley and smaller projects (and major prospective projects, such as Wafi-Golpu), it has also been the main import terminal for the construction of the facilities in the Highlands, MAP 4  Morobe Districts 77 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea including for the Kutubu oil fields, Porgera and more recently the PNG LNG project, based in Hela province. It is the country’s main manufacturing centre (from fish processing to plastic containers, metal work, engineering, milling and bottling) and service centre for Morobe and the wider region, as well as a seat of learning, with the University of Technology, polytechnic, teacher training and other colleges, and research facilities, notably for forestry. The population, as indicated, comes from the local (e.g. Ahi) community and adjoining areas within the province, with the city developing from the traditional villages, such as Butibum, but also reinforced with in-migration from across the Momase and Highlands regions, and in smaller numbers from the New Guinea islands, notably neighbouring West New Britain, and neighbouring Oro province, for which Lae is also the main trading centre, with products such as betel nut shipped constantly to the Lae market. This cosmopolitan mix coexists in the city and its settlements in relative harmony, although certain conflicts do arise, particularly between certain communities, fuelled by factors such as alcohol, but also land issues, high crime levels, and indirectly by the high cost of living in recent years in the city, and fluctuating, but inadequate formal sector job prospects. Menyamya The remote and mountainous district of Menyamya, with its population (recorded at 87,209 in 2011), has long remained one of the most inaccessible and disadvantaged areas in Papua New Guinea (along with its neighbouring vicinities, and tribal affiliates, in Marawaka District in the Eastern Highlands province and the Kaintiba area in Gulf province). Menyamya District has enjoyed a surfaced road to Menyamya station and Aseki in the past, but this has deteriorated, and much of the rest of the district is inaccessible by road, notably in the wet season, when much of the survey was conducted. The multiple airstrips formerly operating in the district are now closed, except Menyamya’s, which only takes charter services. There is some smallholder coffee production in the district, with a small processing factory in Aseki, but poor transport access has undermined this income earning activity, with much produce recorded as un-harvested. During nutrition surveys in the 1980s this was an area with a high prevalence of malnutrition (notably stunting—National Nutrition Survey 1982/3), including goitre amongst 8–11 year old in schools (from lack of iodine) still in the 2000s (WHO Global Database on Iodine Deficiency, 2007). Huon Gulf Huon Gulf is the province’s largest district by area, and stretches nearly the length of the province, including the extensive lowlands and coastal areas in the lower Markham floodplain and southern Morobe coast, but flanked to the south by, and including some of the mountain ranges, notably the Bowotu range, stretching east 78 Appendices to the border with Oro province. Its population (recorded as 77,564 in 2011) also makes it one of the larger districts in the province, although containing no significant towns (with City of Lae effectively as its hub). Much of the district has relatively good accessibility, with most of the population able to reach roads or travel by boat to markets or service centres within a day’s travel, although seasonally sea, road or river travel (e.g. along the turbulent Watut river) can be hazardous. Communities in the Engati area, adjoining Menyama District, are the most isolated and disadvantaged, along with some inland communities in the mountains from the Morobe coast. Economic activities derive from sales of fresh produce, betel nut and some fish to the Lae market, especially, and modest production of cocoa and coconuts from the lower Markham valley and employment on livestock estates and outgrower sales of poultry. There is a little income and employment related to tourism and forestry along the southern coast, around Salamaua and to the east. Bulolo The Bulolo District is the most populous in the province, after Lae city, with 101,568 recorded in the 2011 Census. It includes the Bulolo and Snake River valleys and part of the Watut, and the Waria valley around the former tea growing area of Garaina in the south east. The district includes the larger townships of Wau and Bulolo and the Hidden Valley Gold mine and extensive areas of alluvial gold mining, exploration and panning the river systems. These and the former coffee estates have provided opportunities, even going back to the 1920s, for immigration from other parts of PNG (mostly from Momase and the Highlands provinces). Some coffee is still produced in the district, along with plantation pine (auracaria) timber and associated processing, whilst a variety of fresh produce is grown for market in the cooler upland valleys, especially. Most of the larger valleys (normally) have relatively good access to main centres, markets and basic services, except the remote Waria valley, where the former tea production around Garaina has long ceased. Malnutrition rates (notably stunting) recorded in the district during the 1982–83 National Nutrition Survey were relatively high compared with most of the province, but below the rates in the adjoining, and relatively more inaccessible, Menyama District. Finschafen Finschafen District, which recorded a population of 54,673 in the 2011 Census, comprises a narrow coastal plain and small offshore (Tami) islands, rising up to the high Cromwell mountain range, and its relatively remote highland valleys. The population is spread across the coastal plain, including township of Finschhafen, and the station of Pindiu and multiple mountain villages. Some cash cropping, including coffee and spices, occurred in the past, but poor access, including cessation of rural 79 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea air-services and deterioration of rural roads has undermined opportunities beyond subsistence agricultural away from the coastal strip. The discontinuation of coastal shipping services has also undermined business opportunities and required access by relatively unsafe small craft. Some recent restoration of certain bridges and stretches of access roads may provide the prospects of renewed access to trade and services for some inland rural communities. Markham The Markham District, with a widely dispersed population of 62,495 recorded in the 2011 Census, embraces the main upper portion of the Markham floodplain and a short stretch of the adjoining Ramu valley, but also extends up on the northern flanks into the steep valleys, like the Leron and Wantoap, and high mountains of the western Sarawaget range, merging with the Finisterre range, and in the south into the Kratke range and valleys, like the Wafu. As in most districts, the highest population densities are largely in the restricted valleys, and away from the more malaria affected lowland areas, but poor access to these highland areas has more recently encouraged, as in other remoter districts, outmigration to cities and main access routes. The Markham and upper Ramu valleys have been centres in the past for low intensity ranch-style livestock production, but since the 1980s large tracts have been partly turned over first to commercial sugar and over the past decade also to oil palm, with large concentrations of employees, both local, but also from adjoining provinces (notably Madang, Eastern Highlands and Simbu) also employed long term in these estates and processing facilities. Production of lowland produce for sale into the neighbouring Highlands provinces is prevalent in the Markham valley, as well as to Lae city. Madang Province Madang is divided into six (6) districts (refer to Map 5), of which 2 were selected for the survey, namely Sumkar (including Census Units on Karkar island) and Rai Coast (including a Census Unit on remote Long Island), plus the Madang urban being added, to supplement the selected sample of (rural) districts and CUs. Madang Madang District, including the town had a population of 110,978, as recorded in the 2011 Census. The district covers the town and the lagoon, with its multiple coastal islands, as well as parts of the Gogol valley and eastern end of the coastal Adelbert range. The urban area, which now embraces some of the islands, like Kranket, which are now effectively part of the town, has long been a tourist hub, as well as provincial administrative and trading centre, including for the cocoa and copra trade from the province (especially produce from Karkar island). Latterly the town and adjoining 80 Appendices MAP 5  Madang Districts district have been hub of a growing onshore-based fisheries industry, including local processing (employing mostly young women) and vessel support. It has in recent years also been the administrative base for the Ramu Nickel Company, with periodically large numbers of overseas as well as local staff. The town, which has limited land within the formal township area, with its extensive lakes and inlets, has seen a substantial growth in settlements, particularly housing people from across remoter, including highlands parts of Madang province, plus from across the Momase region, especially East Sepik and other Highlands provinces. Despite poor soils on the limestone coastal plain, agricultural potential is reasonable in the hills and Gogol and other valleys, with access to the urban market for fresh produce able to provide alternatives for more accessible communities to subsistence production. Planation forestry the Gogol (for a chip-mill in town) has ceased and most logging concessions in the district have already exhausted their resource. Sumkar Sumkar, includes the highly populated and agriculturally productive volcanic island of Karkar and its smaller neighbour, Bagabag Island, and also embraces a stretch of less-fertile adjoining limestone coastal plains, rising up into the lower elevation coastal mountain range, the Adelberts. Sumkar’s population was recorded as 84,944 in the 2011 Census, with well over half of that on Karkar and Bagabag islands. Karkar is a major centre for the copra and cocoa industries, with a stream of workboats transporting raw copra to Madang for processing as coconut oil or directly for export, as well as cocoa. The island has some of the country’s few still operating copra and cocoa estates, as well as smallholder production. Some copra and cocoa are also produced along the coastal plain on the mainland, although many of the estates have ceased to operate, and robusta coffee production has long since ceased. Although 81 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea most tourism occurs in the adjoining Madang District, some is centred on lodges in this district. Apart from parts of the Adelbert hills and mountains, access is mostly good along the ‘north coast road’, and on Karkar Island, and mostly by small craft from the islands, although some larger (and generally safer) work boats operate on these routes. Periodic low prices and outbreaks of pest and diseases, notably cocoa pod borer (and Bogia syndrome in coconuts) have affected production and household incomes, which had been relatively high for rural areas. Other cash crops include betel nuts and some surplus fresh produce for sale in Madang and the Highlands markets, although Madang is a far smaller hub than Lae, but it does have significant and growing industries, including tourism, requiring supplies. Rai Coast The Rai Coast District recorded a population of 83,218 in the 2011 Census, largely in the Rai coast rural LLG area within the district. The district covers an extended stretch of relatively straight coastline commencing in the west from flood plains feeding into Astrolabe bay and a progressively rising backdrop to the east, with most of the coast flanked by the Finisterre mountains rising to nearly 4,000 metres, with high population densities in the relatively inaccessible high valleys. The district includes the active volcanic Long island, with its large lake-filled caldera, and the small adjoining Crown Island. The Astrolabe Bay end of the district and into the mountains have high rainfall and a relatively short dry season, whereas in the rain-shadow of the Finisterre (and Sarawaget) ranges to the east, there is a long dry season and agricultural opportunities restrained, with cattle grazing on some of grasslands near Sialum, in Morobe to the east. The western coastal villages have reasonable access by road, when rivers are crossable, and by boat to Madang, but travelling to the remoter eastern parts of the district (including to Long Island, which lies 70 km from the Saidor station, and 130 km from Madang) and inland from the coast involves many hours of travel on foot or by boat, constrained also by the discontinuation of most formal coastal passenger, freight and rural air services. During the wet season, with multiple unbridged rivers, travel is harder and also dangerous by (the predominant) small craft during extended periods of strong winds. The Ramu nickel and cobalt mine, located in the Ramu valley, transports the extracted ore by 100 km slurry pipeline to a preliminary processing plant in the Basimuk Bay along the Rai coast, from where the product is shipped, with the waste dumped, controversially, offshore. Economic opportunities are very limited in most of the district, with a few employed in the port/processing facility, and a little cocoa, copra (including from old estates near Saidor) and betel nut produced on the coast, although, with years of low prices and lack of coastal shipping, the copra trade has largely dried up. Likewise modest vegetable shipments from the inland valleys have dwindled, leaving subsistence agriculture and coastal fishing as almost the only means of livelihood for most of the district’s population. 82 Appendices Appendix 2: Overview of Households Respondents The survey was completed by the female and male in the household who made most financial decisions on behalf of the household. Typically this was the husband and wife. Eighty-seven percent of surveys were completed in households in which both a woman and man were interviewed. Overall, interviews were evenly distributed between women and men. Respondent’s mean age was similar, irrespective of location: 38.63 years for urban respondents and 37.56 years for rural respondents. The mean female age was 35.8 years and the mean male age 39.9 years. As shown in Figure 28, the age distribution for women was slightly younger than that for men. Household Overview Household Size The average household size was 6 members. Household size and composition was similar across rural and urban households. The typical household comprised three adults and three children. There was an even gender balance (52 percent male and 48 percent female). Sources of Income Households reported between 2.3–2.6 sources of income, across rural and urban households. On average, between 2.3–2.4 household members contributed to household income. The principal income earners were in most households the female and male interviewed for the survey. As shown in Table 24, households in rural communities derived most income from the informal sector or self-employment. Figure 28 Respondent Age Distribution 50.0% 40.0% 30.0% 20.0% 10.0% 0.0% <20 20–29 30–39 40–49 50–59 60> Male Female 83 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 24  Principal Source of Household Income Urban Township Rural Formal sector 51.60% 60.20%  7.60% Informal sector/self-employed 33.80% 36.40% 88.10% Other 14.60%  3.40%  4.30% By contrast, in urban and township communities, approximately 50 percent of households derived most income from formal sector employment. Approximately 15 percent of urban households reported the principal source of income from sources other than formal employment or informal sector income. Sources of income were forms of passive income: income from rents and leases, income from investments, pension income. For most households, household income varied over the year. Not surprisingly, given the source of income, a larger percentage of rural households stated income was variable (89.1 percent) than urban households (60.6 percent). Household income tended to be variable across both high and low income earning periods. Rural households were more likely to manage income earned by household members jointly as a household (60.1 percent of households) than urban households (46.7 percent of households). Conversely, urban households that earned business income were more likely to keep the income separate to household income (67.3 percent) than rural households (37.9 percent). As shown in Table 25, sources of income were similar for women and men in rural communities. In urban households, however, men were more likely than women to report formal sector employment. Individual and Household Income Respondents were asked to estimate their average monthly individual and household income from all sources. Income was not verified and has not been used in the Table 25  Principal Source of Individual Income Urban Township Rural Male FEMale Male FEMale Male FEMale Formal sector 52.20% 35.90% 72.10% 27.10% 10.20%  4.30% Informal sector/ 30.90% 53.00% 27.90% 68.80% 84.80% 92.60% self-employed Other 16.90% 11.10%  0.00%  4.10%  4.90%  3.10% 84 Appendices Table 26 Estimated Monthly Income by District Predicted Poverty Level Predicted Poverty Level Highest Middle Lowest Mean Mean Mean Estimated monthly individual earnings $281 $340 $539 Estimated monthly household earnings $366 $410 $691 analysis. As shown in Table 26, respondents’ estimate of their personal monthly income and their household’s monthly income increases as the predicted poverty level of the district reduces. Overall, men and women estimated similar levels of monthly household income (men: $579, women $556). However, it appears respondents may have limited awareness of the income earned by the household. On average respondents estimated they earned $445 per month and their household earned $569 per month. Estimates of individual income earned by urban respondents were approximately 2.3x the estimate of income earned by rural respondents. In rural households men and women estimated similar levels of individual monthly income. However, in urban and township households income estimates by men were higher than estimates by women (refer to Table 27). This is likely to be due to the higher incidence of formal sector employment by men. Disability Four percent of households reported a household member with a serious illness or disability. Information as to the nature of the disability or serious illness was not sought. Sixty percent of household members who had a serious illness or disability were under the age of 18 or 60 years or older. Table 27 Estimated Monthly Individual Income by Gender, Location and Livelihood Group Urban Township Rural Urban Township Rural Informal Informal Informal Sector/ Sector/ Sector/ Formal Self- Formal Self- Self- Male FEMale Male FEMale Male FEMale Sector employed Sector employed employed $1,581 $828 $919 $303 $388 $310 $1,246 $679 $703 $540 $278 85 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 28  Highest level of Education in Household Urban Rural No formal education  0.0%  7.2% Primary  4.9% 57.6% Secondary 35.0% 20.4% Tertiary 60.1% 14.9% Education As shown in Table 28, reported levels of education suggest there may have been a misunderstanding in respect to the meaning of tertiary education. It is possible there was confusion between post-secondary education and senior high school or vocational school. Twenty-two percent (22 percent) of households reported at least one household member having ‘tertiary’ education. It is considered unlikely 60 percent of urban households in Morobe and Madang have at least one household member with post-secondary education. Level of education has not been used for the analysis of financial inclusion or financial capability. Dwelling and Resources Dwelling The assessment of the dwelling was undertaken during enrolment, by a combination of observation and questions. Houses in urban communities were typically larger than houses in rural communities (3.7 rooms compared to 3.0 rooms). Houses in urban communities were constructed primarily from concrete, wood or cement sheet (65 percent of properties), or corrugated iron (17.5 percent of properties). A further 17.5 percent of properties were constructed from traditional or makeshift materials. Most properties (89 percent) had an iron roof. In rural communities most houses were constructed from traditional materials (80 percent) or wood or cement sheet (13.6 percent of properties). The roof was either traditional materials (66 percent) or corrugated iron (30 percent). Between 70 percent–80 percent of the dwellings were assessed by the interviewer undertaking the assessment to be in sound condition and to be weatherproof. Twenty percent (20 percent) of urban dwellings were assessed as being in very good condition. Patterns of land ownership varied. As shown in Table 29, in rural communities households typically either owned the land on which the house was constructed, or owned the land communally. In urban communities approximately 31 percent of land was either owned directly or owned communally. Between 19–47 percent of land in 86 Appendices Table 29 Land Ownership Urban Rural Freehold owned by dwelling owner 20.60% 43.70% Communal 10.90% 44.60% Formal residential settlement 28.50%  2.20% Squatter settlement  2.70%  3.70% Leased land 19.10%  3.20% Other 18.10%  2.60% urban communities appears to be leased. It is not known to what extent the leases are formal. The mode of land ownership for the remaining 18 percent is not known. In rural communities 90 percent of dwellings were owned by the household. In urban communities 46 percent of dwellings were owned by the household, 25 percent were rented, and 28.5 percent were government, institutional or employer provided housing. Services Seventy six percent (76 percent) of rural households had no access to any form of electricity, 10 percent used solar, 6 percent a generator and 7 percent had access to mains electricity. By contrast in urban communities 70 percent of households had mains electricity to the house, 10 percent used solar, 5 percent used a generator and 15 percent had no electricity available. In rural communities nearly all households (98 percent) reported using wood for cooking. In urban communities a wider range of forms of cooking was used: 38 percent of households cooked using wood, 28 percent used LPG gas, 29 percent used electricity and 5 percent used kerosene. The source of water in urban households was mains pipe. In rural households water was sourced from a well, spring or river. Ninety four percent (94 percent) of rural households used a pit latrine toilet, compared to 50 percent in urban households. Fifty percent (50 percent) of urban households used a flush toilet. Resources Nearly all rural households (95 percent) had access to land for food cultivation (either farm land, communal land or a back-yard). By contrast only 33 percent of urban households had access to land to cultivate food. Sixty-seven percent (67 percent) of household did not have access to land to grow food for the household. 87 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 30 Services and Access Urban Township Rural Avge dist. Avge dist. Transport (km) Transport (km) Transport Primary Walk 1.95 Walk 35%,  2.39 Walk school Public Transport 65% Secondary Walk 17.2 Public 25.69 Walk 14% school transport Public transport 76% Private transport 6.5% Health post Walk 1.05 ND 15.32 Walk 76% or hospital Public transport 20% Durables Urban households had, on average two appliances, typically a refrigerator and a TV (or a PC). Rural households typically had no appliances (to be expected given the lack of mains electricity). Services Very few households owned any form of private transport and either walked to services or used public transport. Nine percent (9 percent) of rural households were within 1 hour of the provincial capital (Lae or Madang), 28 percent were between 1–3 hours, and 63 percent of rural households were at least 3 hours travel from the provincial capital (11 percent of households were more than 9 hours from the provincial capital). Modes of transport to services and rural distance to services are shown in Table 30. Appendix 3: Sampling Dr. Gibson prepared two sampling notes for the financial capability survey. The first sampling note was drafted for the national survey. The second sampling note was drafted following the reduction in scope of the study to Morobe and Madang provinces. Both sampling notes are included as the re-weighting undertaken by Dr. Gibson which is discussed in the second sampling note, must be read in the context of the initial sampling for the national study. 88 Appendices Appendix 3a: Sampling Notes for PNG National Financial Capability Survey John Gibson. 13 November 2013 Summary Sample selection is based on a stratified four-stage random sample design, with preliminary counts from the 2011 Census of Population providing the sampling frame. Stratifying according to female literacy and predicted poverty, the first two stages selected 19 districts from seven provinces, with probability proportional to estimated size. The provinces include three from the Highlands (Western Highlands, Jiwaka, Eastern Highlands), two from Momase (Madang and Morobe), and one each from the New Guinea Islands (East New Britain) and Papuan (Central) regions. Additionally, the National Capital District (NCD) is a separate survey strata. Within each district, five Census Units (CUs) have been selected with probability proportional to estimated size (PPeS). Within each CU, ten households are to be selected by the interview teams, using circular systematic sampling. This will yield an overall sample of 50 households per district and 950 households in total. A further 150 households are to be surveyed from the National Capital District (NCD), with a target of six households per CU, in 25 CUs that have been selected with a PPeS approach. Combining the strata, a total of 1100 households are to be surveyed, and in each household an adult male and adult female are to be interviewed (these should be the primary financial decision-makers in cases where there is more than one adult of each gender). Since not all households will have both an adult male and an adult female, the final sample size will be less than 2200 but should be approximately the sample of 2000 that survey organizers have indicated is the maximum that is feasible and affordable. Survey weights are required to provide nationally representative estimates. These weights reflect deviations between the actual number of households surveyed per CU and the target, differences between the preliminary Census count of households in each CU and the number established during the listing phase of the survey, and between the preliminary and final Census counts for districts and provinces. The oversampling of the NCD is also accounted for with the weights. Surveying in PNG can sometimes be affected by difficulties in accessing locations due to tribal fighting and other security risks, and more generally due to infrastructure problems. The survey is expected to be in the field from early 2014 and has only a few months before fieldwork is meant to be complete. Consequently it may not be possible to simply wait for an improvement in the circumstances that block access to locations. In the case of localized disruptions, choosing an alternate CU from the same district is sufficient, and a list of 19 replacement CUs has been selected (one per 89 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea selected district) using a PPeS draw. Occasionally the disruption is more widespread and may block access to an entire district and choosing a replacement in these cases is a more serious matter (it is varying almost 5 percent of the sample). A ‘nearest neighbour’ matching approach is proposed, where an unselected district from within the seven selected provinces with the closest probability of selection to that of the blocked district is chosen, using a model with predicted poverty and female literacy as the matching covariates. Any replacement district chosen using this approach will be identified at the time, by the sampling consultant, and only following the approval of the task manager and steering committee for the survey. Sample Constraints The discussions in Port Moresby with survey organizers indicated that the survey unit of observation was an individual (adult), that it was desirable to have equal numbers of men and women, and that the available budget was likely to allow a maximum of 2000 observations. The basis of the calculation of this maximum sample that could be afforded was not considered, although there are obviously cheaper and dearer ways to survey 2000 respondents. The survey designer also indicated that spreading these 2000 observations over a full-scale national survey would be impractical, not only due to the high fixed costs of only a small sample size in each area but also because of the limits to supervision capacity, leading to quality control problems, and the likelihood of disruption in one or more sites from infrastructure problems, tribal fighting and other security risks. On the other hand it is important that the sample not degenerate into a limited number of case studies, which might be dismissed as unrepresentative, and which would not be a robust basis on which to develop financial inclusion policies. Sampling Individuals versus Households Although the ultimate interest is in a gender-stratified sample of individuals, it is an inefficient use of survey funds to go to 2000 different households to survey 2000 individuals. Such an approach also will preclude the use of some empirical methods for the analysis once the unit record data are available. While there is potentially a gain in statistical efficiency (precision) because a sample built up by having one person selected from each of 2000 different households will have more variation than a sample of 1000 households, each supplying two respondents, this is not sufficient advantage to recommend the design of one-individual-per-household, for the following reasons: ■■ Surveys on financial topics typically have high rates of refusal, so with a one-respondent-per-household rule, it may be that any differences that emerge between males and females are due to differential non-response. For 90 Appendices example, females in richer households may be more likely to respond and males in poorer households more likely to respond, so what appears to be a male-female difference in the data is in fact driven by an income difference in survey compliance. This lack of ‘balance’ in the gender sub-samples especially matters to this survey since there is no plan to collect detailed household- level data on incomes or consumption (interviews are expected to be only one-hour per respondent) which could otherwise be used as covariates to control for these threats to validity posed by differential non-compliance. ■■ Surveys of household wealth and financial management often find discrepancies between what males and females in the same household report on ostensibly the same matters (see e.g. various studies from the Household Income and Labour Dynamics in Australia survey). This is an important behaviour to observe since it can be informative about the extent to which financial information, decisions and risks are pooled within the household, and it is impossible to observe this with only one respondent per household. ■■ The lack of detailed household-level data on consumption, income or wealth makes it possible that statistical analyses of the data are susceptible to omitted-variable bias. Such bias occurs when variables not captured by the survey are relevant to the behaviour under study and are correlated with policy-relevant variables that the survey captures and that analysts use to try to explain behaviour. For example, there may be more joint financial decision making in richer households, but these are also households where women’s education is typically higher, so this difference in financial decision-making could wrongly be attributed to an effect of women’s education rather than to an effect of unobserved household wealth. One partial solution to this problem is use household fixed effects (dummy variables) as follows: yi j = a j + Di + xij + e ij where yi j is the outcome of interest for person i in household j, a j is a household-specific fixed effect that captures average differences between households, Di is a dummy variable to indicate gender of the respondent, xi j are control variables such as education or age that can vary between individuals in a household and e ij is a random error. In other words, after controlling for observable covariates of individuals and for differences between each household (the a j can be considered household-specific conditional means), is there a systematic difference in behaviour y between men and women? However this estimation framework relies on having more than one respondent per household. 91 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea ■■ It is only in coastal PNG that rural villages are spatially concentrated (so there is only a low marginal cost to go from one dwelling to a neighbour in order to survey an extra individual). In the Highlands, many rural Census Units are simply clusters of scattered hamlets, so it is expensive in time and energy to reach a particular household, so once there it makes sense to interview more than one respondent. These advantages of obtaining 2000 respondents by (ideally) surveying one man and one woman per household in 1000 households make this the preferred design and the selection of Census Units outlined below has proceeded on this basis. But survey funders and implementers should be aware of two implications of this design: (a) it requires a minimum of two person teams, so that men can be interviewed by men and women by women, and (b) some redundancy will be needed in the number of households surveyed since it will not be possible to always find one adult male and one adult female per household. The average number of adults (age 18+) per household ranges from 2.4 in Manus to 3.5 in Southern Highlands and NCD, so most households should have at least two adults and these are likely to be a man and a woman. But households with widows and widowers, households where one adult is away temporarily (e.g. at work camps), and units of polygamous households where each wife has an independent dwelling and a husband rotates between these, will all be expected to yield only one adult respondent. Moreover, the gendered nature of many rural tasks means that often it will not be possible to find a male and a female from the same household together at the same time since they are conducting different tasks in different places. The sample design described below allows for a fieldwork schedule where teams are in a CU for approximately 3–4 days, so there is time for return visits to those households where one adult was absent on the initial visit, although interviewers are often unwilling to make these extra efforts. Therefore the sample is designed for 1100 households, assuming that for approximately 100 households it will not be possible to survey both a man and a woman. These ‘unbalanced’ households are still useful in the analysis so this sample redundancy does not imply inefficient use of survey funds. Target Number of Households per Census Unit Moving up a sampling level, from the question of how many individuals per household to survey, the next question is how many households per Census Unit (CU) to survey. A discussion of the related issues of how to select households in each CU is (that is, the listing and selection approach) and how to plan for refusals and other non-response is covered later. It is advantageous to require the teams to survey a fixed number of households per CU, so that a standardized fieldwork plan can be adhered to, which reduces interviewer discretion (so that lazy interviewers cannot claim that the CU was smaller than expected so they reduced their quota of interviews accordingly, in a design where the target number of households varies with CU size). 92 Appendices In general, clustered survey designs are less efficient, the more highly correlated within local areas are the behaviours under study. Conversely, where there is a large idiosyncratic component, even designs where up to 20 households are surveyed in each enumeration area (EA) can still see the 20th household providing some new information. In the case of financial literacy or competency, most studies find that this varies with experience of various financial transactions and with education. Both of these factors are likely to be highly correlated within CUs since they reflect the proximity to financial institutions (which is common for everyone in the same CU) and education is highly related to wealth, which affects residential sorting. It is therefore best to allocate the sample of 1100 households over more CUs, with fewer households surveyed per CU. This design is especially recommended because many CUs in PNG are quite small, averaging just 50 households. In the 1996 household survey, out of 80 selected CUs (outside of the NCD), three (4 percent) were too small to enable the target of 12 households to be surveyed, and so had to be combined with neighbouring CUs, with listing and household selection then carried out on the pseudo-CU created from the joining of the two neighbours. This is a complicated process which it is best to avoid at the outset. The recommendation is that outside of the NCD, ten (10) households per CU should be selected to survey. For the NCD sample (the reason for this being a separate strata are discussed below) the target should be six households per Census Unit. The reason for the different targets is that fieldwork fixed costs (e.g., transportation to the CU) are lower in an urban area like the NCD, so the sampling efficiency advantages of a smaller cluster size outweigh the budgetary advantages of a larger cluster size. Moreover, it may be possible to avoid a listing operation in the selected CUs for the NCD since the NSO mapping of those CUs is relatively recent, allowing selection of dwellings to be done from paper maps. A supplement (or fall-back) to relying on the NSO maps of NCD Census Units is to use Google Maps aerial images, which have sufficiently high resolution for NCD that it allows easy identification of individual dwellings, even in a densely populated urban village like Hanuabada. Consequently, the fixed cost of listing a CU should matter less in NCD, so there is a smaller time and cost disadvantage to offset the statistical efficiency advantages of spreading a given sample size over more CUs. Finally, it is likely that surveying individual dwellings is more difficult in the NCD, because all adults are absent during the day, because night time surveying is infeasible due to security issues, and because of difficulty in gaining access to guarded residences. Therefore more “reserve” households should be allowed for in each CU, which dictates having a smaller number of households initially targeted for selection (allowing more to be kept for reserves). Stratification and Stage One and Two Selections Stratification allows us to use prior information to structure a sample so that it reflects the range of conditions in the population, rather than relying on this occurring 93 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea by chance (which it would only do in the right proportions if we had the luxury of drawing repeated samples). Another reason to stratify is to introduce uneven sampling weights, if it is desirable to over-sample a particular area. Along these lines, at the outset a decision was made to divide the sample into two main strata: (i) the National Capital District (NCD), and (ii) all other areas of the country. This was to ensure that the NCD was included in the sample, and given almost four-times the sample size than its population would justify, ensuring sufficient sample size in the event of any follow- up survey. While it would be ideal for a follow-up survey to have national coverage, the reality is that it is often easier to obtain (donor) funding for new, baseline surveys than it is for on-going follow-up surveys. Consequently, the logistical and financial constraints on any future follow-up survey may be such that it is restricted to the NCD, so ensuring sufficient sample size in that region at baseline is useful. The different sampling fractions for the NCD and the rest of the country will be controlled for by using weights (expansion factors) when forming national estimates. An additional reason for forcing the NCD into the sample is that this area has the most favourable conditions in the country, in terms of access to financial services and levels of education that should support informed use of financial services. Hence, if the survey identifies major shortfalls that occur in these more ideal conditions in the NCD it is potentially revealing about the scope of the task that financial inclusion interventions will require in more challenging regions of the country. Finally, the experience in the NCD provides a natural comparison point which PNG officials (both elected and appointed) should relate to, given the unfortunate reality that many of them do not have recent first-hand experience of conditions in more remote districts but they do have first-hand experience of the NCD. Stratification also was applied to the non-NCD sample in order to improve the precision of sample estimates. All provinces were sorted according to the female adult literacy rate, prior to the first-stage PPeS selection of provinces. The first reason for using female literacy as the stratifying variable is that gender differences in financial competency, in the degree of discrimination against women in access to and use of financial services, and in household decision-making that leads to gendered effects in the use of money, ultimately rely on variations in female bargaining power, which is plausibly higher, the more educated are females (relative to males). The second reason is that prior studies of financial literacy of Pacific Islanders find that greater formal education is the characteristic that is most predictive of an individual having superior financial literacy. In PNG, the provinces with highest female literacy rates also tend to have the highest overall education levels, so stratifying provinces by female literacy rates ensures that the survey will cover the full range of conditions for these two factors that are likely to be closely associated with the financial competency behaviours under study. The available data on female adult literacy rates came from the 2000 Census, since final estimates from 2011 were not available. This reliance on prior estimates should 94 Appendices not overly matter since the spatial pattern of female literacy likely changes only slowly. One implication of using data from the 2000 Census is that it was necessary to form estimates for the two new provinces of Jiwaka and Hela (which did not exist in 2000), and also to revise estimates for the Southern Highlands and Western Highlands, which are the two existing provinces that the new provinces have been carved out of. The new provinces were formed by reallocating three districts from each of the existing provinces, so district-level literacy rates were aggregated (weighing by the number of adult females) to form estimates corresponding to the new and revised provinces. The resulting ranking of provinces is shown in Table 31, with the female literacy rate ranging from 29.5 percent in Enga to 83.7 percent in Manus (and 88.7 percent in the NCD, which is in a separate stratum). This variable clearly captures some of the considerable heterogeneity that occurs between provinces in PNG, which it is important to have reflected in the sample that is selected. In order to select from amongst these ranked provinces, four values are required: a) The total number of households from the 2011 preliminary Census counts (1,367,029 excluding the households in the NCD). b) The target number of provinces, set here as seven (7) to ensure that a range of circumstances are covered by the sample while not spreading fieldwork too thinly. c) The ratio of (a) to (b), which gives the sampling interval of 195,290. Essentially the province containing every 195,290th household is selected after a random start, where this naturally gives higher odds of selection for a province with more households (which is the basis of PPeS selection). d) The random starting value, obtained by multiplying the sampling interval by a random number. The random number which was generated in Excel was 0.9121, so the random starting value is 0.9121 3 195,290 = 178,124. In other words, the province containing the 178,124th household was the first selected, and then skipping down another 195,290 households, to find the next selected province, which is the one containing the 178,124 + 195,290 = 373,414th household, and so forth. The seven selected provinces outside of the NCD contain 38 districts (out of 87 districts in all of PNG) and the second stage sampling involved selecting 19 of these districts. The first stratifying variable used was the female literacy rate (at district level), with each district put into one of three groups: ■■ Adult literacy rate for females is ≤ 33 percent (n = 12) ■■ Adult literacy rate for females ranges from 33.1 percent–66.9 percent (n = 14) ■■ Adult literacy rate is ≥ 67 percent (n = 12) 95 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 31  First Stage Selection of Provinces Female Literacy Province (%) # of Districts # Households Cumulative Set 04. National Capital District 88.7 1 57741 08. Enga 29.5 5 85012 85012 21. Hela 30.5 3 65309 150321 09. Western Highlands 32.0 4 78560 228881 1 07. Southern Highlands 33.0 5 94987 323868 22. Jiwaka 33.4 3 68835 392703 2 10. Chimbu 34.5 6 79888 472591 15. West Sepik 36.4 4 45863 518454 11. Eastern Highlands 36.5 8 133425 651879 3 14. East Sepik 46.0 6 89952 741831 13. Madang 48.8 6 94176 836007 4 02. Gulf 50.9 2 24331 860338 12. Morobe 57.3 9 132993 993331 5 01. Western 66.1 3 34573 1027904 06. Northern 66.2 2 33704 1061608 19. West New Britain 66.9 2 49077 1110685 03. Central 68.6 4 46662 1157347 6 20. AR Bougainville 75.0 3 47888 1205235 17. New Ireland 75.8 2 34422 1239657 05. Milne Bay 76.2 4 57626 1297283 18. East New Britain 80.7 4 58517 1355800 7 16. Manus 83.7 1 11229 1367029 Total (a) 87 1367029 Target number of provinces to select (b) 7 Sampling interval (a/b) 195290 The second stratifying variable was based on a ranking of districts according to their predicted poverty rate that came from a poverty mapping exercise using the 1996 PNG Household Survey and the 1990 Census, their ranking according to the “development index” calculated by the National Economic and Fiscal Commission, and their ranking under the “disadvantage index” calculated in the District Planning handbook that the Australian National University produced for the Department of 96 Appendices Table 32 Cross-Tabulation of Female Literacy Group and Poverty/Disadvantage Group Amongst Districts Within the Selected Seven Provinces District-level disadvantage and poverty Highest Medium Lowest Poverty Poverty Poverty Lowest (≤3%) 4 3  5 Female literacy Medium (33.1–66.%) 2 2 10 group Highest (≥67%) 0 3  9 42 National Planning and Monitoring. The rankings of districts according to these three (closely overlapping) measures are discussed in Gibson et al. (2005), focussing on 43 the 20 most disadvantaged districts under each ranking. A group of 13 districts are ranked worst according to both the predicted poverty rate and either or both of the other two indexes. These are locations within PNG that are widely agreed upon to be disadvantaged and six of these districts are from the seven provinces selected for the survey so it is important that the survey captures conditions in some of these areas. The next group of 23 less disadvantaged districts have at least one of the three indexes ranking them amongst the 20 most disadvantaged districts, so these are areas for which there is reasonable agreement that people living there face considerable difficulties. Amongst this second group of 23, there are 8 districts that come from the seven selected provinces and it is important for this level of disadvantage to also be represented in the final sample. Finally, the remaining 51 districts, which include urban areas such as Lae, are not typically considered areas of major disadvantage or poverty in PNG and there are 24 such districts within the seven provinces selected. The cross-tabulation of the two stratifying variables in Table 32 shows one empty cell, for the combination of highest poverty and highest female literacy. All other strata have at least two districts, supporting use of a ‘nearest neighbour’ approach to selecting a replacement district in exceptional cases where it is impossible to conduct fieldwork in the original selected district. 42  The disadvantage index is a composite of five indicators; each ranging from 1–5, measuring land potential, agricultural pressure, access to services, income from agriculture, and child malnutrition. The index thus ranges from 5 (most disadvantaged) to 25 (least disadvantaged), and it is ranks of the index that are used here. 43   Gibson, J., Datt, G., Allan, B., Hwang, V., Bourke, M., and Parajuli, D. (2005). “Mapping poverty in rural Papua New Guinea.” Pacific Economic Bulletin, 20(1): 14–29. 97 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 33 Second stage selection of districts in the seven proviinces # Name Province Pov_group Lit_group households Cumulative Selected GOILALA 3 1 1 6366 6366  1 OBURA/WONENARA 11 1 1 8363 14729 MIDDLE RAMU 13 1 1 15026 29755 JIMI 22 1 2 14465 44220  2 BOGIA 13 1 2 15007 59227 RAI COAST 13 1 2 13595 72822  3 TAMBUL/NEBILYER 9 2 1 16205 89027 MENYAMYA 12 2 1 16869 105896  4 USINO BUNDI 13 2 1 11706 117602 KABWUM 12 2 2 9051 126653 POMIO 18 2 2 12910 139563  5 RIGO 3 2 3 9800 149363 ABAU 3 2 3 9236 158599 Finschafen 12 2 3 11120 169719  6 MUL/BAIYER 9 3 1 19412 189131 DEI 9 3 1 16960 206091  7 HENGANOFI 11 3 1 14970 221061 OKAPA 11 3 1 16578 237639  8 LUFA 11 3 1 14892 252531 MT Hagen 9 3 2 26022 278553  9 UNGGAI/BENNA 11 3 2 15722 294275 GOROKA 11 3 2 22405 316680 10 ASARO/WATABUNG 11 3 2 12857 329537 11 KAINANTU RURAL 11 3 2 27650 357187 BULOLO 12 3 2 20639 377826 12 TEWAE/SIASSI 12 3 2 11124 388950 MARKHAM 12 3 2 13832 402782 13 NORTH WAGHI 22 3 2 16555 419337 ANGLIMP/SOUTH WAGHI 22 3 2 37851 457188 14 KAIRUKU/HIRI 3 3 3 21297 478485 15 NAWAE 12 3 3 9260 487745 LAE 12 3 3 23661 511406 16 98 Appendices Table 33 Second stage selection of districts in the seven proviinces (continued) # Name Province Pov_group Lit_group households Cumulative Selected HUON 12 3 3 17515 528921 17 MADANG 13 3 3 22020 550941 SUMKAR 13 3 3 16980 567921 18 GAZELLE 18 3 3 23046 590967 19 KOKOPO 18 3 3 15609 606576 RABAUL 18 3 3 7076 613652 The 38 districts in the seven selected provinces are listed in Table 33, sorted according to their stratifying variables, and then according to province code. The number of households in each district, the cumulative number of households and the order of selection are also shown. The selection of 19 districts from amongst the 38 available in the seven provinces followed the same approach that was used for the first-stage selection. Specifically, four values were required: a) The total number of households in all 38 districts from the 2011 preliminary Census counts (613,652). b) The target number of districts (19). c) The ratio of (a) to (b) gave the sampling interval of 32,297, so the district containing every 32,297th household is selected after a random start (so more populous districts are more likely to be in the sample under this PPeS selection). d) The random starting value, obtained by multiplying the sampling interval by the random number (0.1604 generated in Excel) was 5179, where the district containing this household was the first district in the stratified list—Goilala. Thereafter, skipping down another 32,297 households, to find the next selected district and so forth. The result of this selection process is summarized in Table 34, in terms of the coverage of the various strata. The range of conditions existing in the selected seven provinces are well covered by the selected districts, with every feasible cell (noting there are no districts in these seven provinces in the high literacy-high poverty cell) having at least one representative. Thus, even though only 19 districts out of the 87 districts in Papua New Guinea have been selected into the sample, these cover the varying range of circumstances that households face. 99 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 34 Cross-Tabulation of Female Literacy Group and Poverty/Disadvantage Group Amongst Selected Districts District-level disadvantage and poverty Highest Medium Lowest Poverty Poverty Poverty Lowest (≤33%) 2 1 2 Female literacy Medium (33.1–66.9%) 1 1 6 group Highest (≥67%) 0 1 5 In terms of the provincial break-down of the selected districts, there are two from each of Central, Western Highlands, Jiwaka, Madang and East New Britain, three from Eastern Highlands and six from Morobe. It may appear surprising that one province supplies six of the selected districts, but this reflects the size and heterogeneity of this particular province, which has the largest number of districts of any province, and the spread of these over five different strata. This spread is symptomatic of the intra- province differences that exist in PNG, which is also exemplified by Central province, where the two selected districts neighbour each other but come from opposite ends of the stratified ranks (Goilala is in the highest poverty-lowest literacy class and Kairuku/Hiri is in the lowest poverty-highest literacy class). Moreover, it is important to emphasize that the aim of the sampling is not to produce estimates that are representative for a particular province. Instead the aim is to represent the range of conditions found throughout PNG so as to build up representative national estimates. Thus, even with six districts selected from Morobe, the results are not representative for Morobe since three districts are not selected and since the six districts chosen were not meant to be representative at Provincial level. Likewise, even with the same inclusion rate of two-thirds, the results for Jiwaka where two out of the three districts in that province are in the sample, will not be expected to be representative of that province. Instead the selected districts are representing combinations of female literacy and overall disadvantage or poverty that occur in other areas of Papua New Guinea as well, and that are expected to lead to variation in (gendered) financial competency. Stage Three Selections of Census Units The logistics of survey organization are likely to be simplified if the same number of Census Units is targeted to be surveyed in each of the selected districts. In this way, it is possible to organize more standard team sizes and workloads, and it removes 100 Appendices discretion from field staff, who may exploit any discretion in the choice of how many CUs to survey in each district by shirking. The recommendation is to survey five Census Units per district, which would provide a workload of approximately 12–15 days per team, under an assumption that teams can interview 3–4 households per day, and depending on the time needed for listing in each CU and for movement of the teams between Census Units in the same district. One possible concern with a fixed quota of CUs per district is the uneven size of districts, which at the extremes range from 6,366 households in Goilala to 37,851 households in Anglimp/South Waghi (both based on preliminary Census counts). This concern would matter if districts had been selected with equal probability (say, using simple random sampling), since Census Units in small districts then have higher odds of being selected. But in fact, the selection is based on probability proportional to size, and so the odds of a small district like Goilala being selected are much lower than are the odds for a large district like Anglimp/South Waghi. Because the sample weights are the inverse of the selection probabilities, the CUs selected in Goilala already have a large weight (because they represent a large number of households in other small districts in PNG). Selecting fewer CUs from Goilala because of its small size would make the weight for the remaining CUs in that district even larger, causing statistical and economic inefficiency (extreme weights make the results sensitive to the value for particular observations, and indicate a poor spatial budget allocation since money is spent surveying in some CUs that have little impact on the overall weighted results). This same logic of probability-proportion-to-size selection also explains why it is possible to survey the same number of households (10) in Census Units that can be quite variable in size, ranging from fewer than 20 households to sometimes several hundred. Since larger Census Units already have a higher probability of selection, there is no need to survey more households from a larger CU than a smaller CU, and to do so would be to introduce unequal weights. The advantage of a fixed quota of households to interview per CU is that it reduces incentives for interviewer shirking because it removes any discretion over sample selection. The selected Census Units for the 19 districts are listed in Table 35, along with the names of the district, of the Local Level Government (LLG) and of the Ward that the Census Unit is part of. The preliminary count of households in each Census Unit is also listed, and it is this preliminary count which influences the selection probability. When updated estimates of the size of each selected CU are available from the listing done by the interview teams, the final survey weights can be calculated, which take into account any difference between preliminary and final counts. It is for this reason, along with ensuring a representative sample of households from each CU rather than an unrepresentative convenience sample, that a listing operation is required. 101 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 35 Selected census units (outside of NCD) # of District Local Level HOUSE- Province District # Government Ward Census Unit HOLDS Central (3) Goilala 2 04.Guari Rural 03.Rupila 004.Enaugagave 42 Central (3) Goilala 2 05.Tapini Rural 05.Kataipa 007.Kariaritsi 5 008.Kataipa 14 Central (3) Goilala 2 06.Woitape Rural 02.Chirime Valley 014.Kaugeri 142 Central (3) Goilala 2 06.Woitape Rural 04.Fane 037.Old Gutiva Vosa 40 & Nesa Central (3) Goilala 2 06.Woitape Rural 07.Ononge 018.Sigufe 36 Central (3) Kairuku/Hiri 3 07.Hiri Rural 08.Barakau 006.Rabuka 46 Central (3) Kairuku/Hiri 3 07.Hiri Rural 18.Boteka 424.Siraka Settlement 138 Central (3) Kairuku/Hiri 3 08.Kairuku Rural 16.Malati 508.Lolorua Sawmill 47 Central (3) Kairuku/Hiri 3 10.Mekeo Kuni 02.Veifa’a 004.Veifa’a 394 Rural Central (3) Kairuku/Hiri 3 10.Mekeo Kuni 16.Upper Kuni 003.Deva Deva #1 44 Rural Western Highlands (9) Dei 2 03.Dei Rural 05.Gumanch.2 005.Gumanch.2 99 Western Highlands (9) Dei 2 03.Dei Rural 14.Kamund 001.Kamund 119 Western Highlands (9) Dei 2 14.Kotna Rural 17.Mala 1 030.Kram 52 Western Highlands (9) Dei 2 14.Kotna Rural 32.Kinjibi 015.Oldor 56 Western Highlands (9) Dei 2 14.Kotna Rural 48.Rulna 061.Rum 39 Western Highlands (9) Mt Hagen 3 04.Mt Hagen Rural 03.Kelua 2 018.Puntibulg 100 Western Highlands (9) Mt Hagen 3 04.Mt Hagen Rural 09.Tiling 033.Wagapil 91 Western Highlands (9) Mt Hagen 3 04.Mt Hagen Rural 22.Pulgimp 026.Keipilkakona 67 Western Highlands (9) Mt Hagen 3 04.Mt Hagen Rural 32.Wimbuka 401.Ambra 97 Western Highlands (9) Mt Hagen 3 05.Mt Hagen 83.Mt.Hagen Town 031.Kerebug 67 Urban Residential Area Jiwaka (22) Anglimp/ 1 01.Anglimp Rural 15.Kutibulg 1 009.Nguakona 87 South Waghi Jiwaka (22) Anglimp/ 1 01.Anglimp Rural 27.Kindeng 1 404.Kindeng Dpi 46 South Waghi Jiwaka (22) Anglimp/ 1 02.South Waghi 08.Kungar 1 016.Pepik 88 South Waghi Rural Jiwaka (22) Anglimp/ 1 02.South Waghi 20.Ngunba Tsents 025.Gun 101 South Waghi Rural Jiwaka (22) Anglimp/ 1 02.South Waghi 48.Kia 006.Rutban 42 South Waghi Rural Jiwaka (22) Jimi 4 06.Jimi Rural 05.Kwima 009.Deka 131 Jiwaka (22) Jimi 4 06.Jimi Rural 24.Maikmol 007.Maikmol 151 Jiwaka (22) Jimi 4 06.Jimi Rural 32.Tabibuga 015.Dapaka 95 Jiwaka (22) Jimi 4 07.Kol Rural 03.Kilmin 019.Kilmin 56 Jiwaka (22) Jimi 4 07.Kol Rural 19.Kouila 015.Ziki 111 Eastern Highlands (11) Asaro/Watabung 1 01.Watabung Rural 01.Mangiro 037.Mangiro 49 (Daulo) 102 Appendices Table 35 Selected census units (outside of NCD) (continued) # of District Local Level HOUSE- Province District # Government Ward Census Unit HOLDS Eastern Highlands (11) Asaro/Watabung 1 12.Lower Asaro 01.Mando/ 100.Mambanuhaloka 37 (Daulo) Rural Yamaiufa Eastern Highlands (11) Asaro/Watabung 1 12.Lower Asaro 03.Asaro No. 1 403.Obiaka Plantation 23 (Daulo) Rural Eastern Highlands (11) Asaro/Watabung 1 12.Lower Asaro 07.Lunumbeyuho 094.Kofiko 18 (Daulo) Rural Eastern Highlands (11) Asaro/Watabung 1 13.Upper Asaro 02.Kombiangu/ 401.Kambiagwe 170 (Daulo) Rural Amaiufa Eastern Highlands (11) Goroka 5 02.Gahuku Rural 03.Kami-Seigu 126.Hauslain 44 Eastern Highlands (11) Goroka 5 02.Gahuku Rural 05.Fimito 070.Ekepoka 83 Eastern Highlands (11) Goroka 5 03.Goroka Urban 80.Goroka Urban 014.Goroka 58 Eastern Highlands (11) Goroka 5 14.Mimanalo Rural 01.Zomaga 405.Six Mile 61 Eastern Highlands (11) Goroka 5 14.Mimanalo Rural 04.Kabiufa No.2 053.Negemoka 158 Eastern Highlands (11) Okapa 7 10.East Okapa 01.Purosa 036.Kakuoti 59 Rural Eastern Highlands (11) Okapa 7 10.East Okapa 08.Yasubi 036.Waniganodo 153 Rural Eastern Highlands (11) Okapa 7 10.East Okapa 12.Ofafina 006.Famia 156 Rural Eastern Highlands (11) Okapa 7 22.West Okapa 03.Wayoepa 034.Wayoepa No 1 172 Rural Eastern Highlands (11) Okapa 7 22.West Okapa 10.Amuraisa 023.Foretu 45 Rural Morobe (12) Bulolo 1 01.Mumeng Rural 17.Kumalu 007.Mumengtain 95 Morobe (12) Bulolo 1 03.Watut Rural 02.Hawata 007.Hikiawa 127 Morobe (12) Bulolo 1 04.Wau/Bulolo 80.Bulolo Urban 008.Forestry 51 Urban Compound Morobe (12) Bulolo 1 05.Wau Rural 04.4 Mile/Nami 411.Wara Kalap 55 Morobe (12) Bulolo 1 29.Buang Rural 04.Moniau 001.Moniau 136 Morobe (12) Finschafen 2 06.Hube Rural 30.Genna 012.Genna 71 Morobe (12) Finschafen 2 07.Kotte Rural 09.Siki 024.Lecko 44 Morobe (12) Finschafen 2 08.Yabim Mape 14.Macwaneng 052.Ziao 66 Rural Morobe (12) Finschafen 2 32.Burum Kwat 08.Sagiro 002.Dubi 45 Morobe (12) Finschafen 2 33.Finschafen 01.Timbulim/ 507.Brown Hospital 107 Urban Tamuc Morobe (12) Huon 3 09.Morobe Rural 03.Miama 009.Miama Village 145 Morobe (12) Huon 3 10.Salamaua Rural 08.Salus 019.Salus 123 Morobe (12) Huon 3 11.Wampar Rural 08.Labubutu 009.Labubutu 221 Morobe (12) Huon 3 11.Wampar Rural 13.Busanim 416.Sepik Settlement 93 Morobe (12) Huon 3 11.Wampar Rural 18.Naromangki 436.Erap C/School 16 Morobe (12) Lae 5 15.Ahi Rural 82.Lae City 640.Kamkumung 202 Market 103 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 35 Selected census units (outside of NCD) (continued) # of District Local Level HOUSE- Province District # Government Ward Census Unit HOLDS Morobe (12) Lae 5 15.Ahi Rural 82.Lae City 674.Busurum 258 Settlement Morobe (12) Lae 5 16.Lae Urban 82.Lae City 010.Quail Crest 146 Morobe (12) Lae 5 16.Lae Urban 82.Lae City 110.Bumbu Police 235 Barracks Morobe (12) Lae 5 16.Lae Urban 82.Lae City 178.Tent City 130 Morobe (12) Markham 6 18.Umi/Atzera 01.Ragiampun 501.Umi/Atzera LG 28 Rural Station Morobe (12) Markham 6 18.Umi/Atzera 09.Yanuf 008.Yanuf 39 Rural Morobe (12) Markham 6 18.Umi/Atzera 21.Sauruan 018.Sauruan 239 Rural Morobe (12) Markham 6 19.Wantoat/Leron 04.Arawek 001.Arawek 113 Rural Morobe (12) Markham 6 19.Wantoat/Leron 20.Ngariawang 005.Wangat Sayang 41 Rural Morobe (12) Menyamya 7 21.Kome Rural 04.Hengiapa 017.Sengiapa/ 98 Hengiapa Morobe (12) Menyamya 7 21.Kome Rural 12.Kenali 011.Epike 148 Morobe (12) Menyamya 7 22.Wapi Rural 09.Mabukapu 014.Hazoi 88 Morobe (12) Menyamya 7 30.Kapao Rural 07.Okaneiwa 004.Okanaiwa (4) 34 Morobe (12) Menyamya 7 34.Nanima Kariba 19.Poiyu 017.Poiyu 61 Madang (13) Rai Coast 4 10.Astrolabe Bay 03.Bang 005.Gur 29 Rural Madang (13) Rai Coast 4 10.Astrolabe Bay 14.Ato 019.Ato Settlement 123 Rural Madang (13) Rai Coast 4 12.Rai Coast Rural 04.Kepoiak 028.Pisangana 43 Madang (13) Rai Coast 4 12.Rai Coast Rural 17.Lamtub 029.Singor 42 Madang (13) Rai Coast 4 12.Rai Coast Rural 35.Bok 001.Bok 80 Madang (13) Sumkar 5 13.Karkar Rural 04.Kaviak 014.Kinim 101 Madang (13) Sumkar 5 13.Karkar Rural 15.Muluk 031.Kurubek 109 Madang (13) Sumkar 5 13.Karkar Rural 26.Marup 020.Marup 1 175 Madang (13) Sumkar 5 14.Sumgilbar Rural 03.Murukanam 009.Murukanam 345 Madang (13) Sumkar 5 14.Sumgilbar Rural 17.Kudas 015.Matukar 86 East New Britain (18) Gazelle 1 01.Central Gazelle 06.Vunagogo 008.Vunagogomor 87 Rural East New Britain (18) Gazelle 1 02.Inland Baining 08.Liaga 001.Liaga 86 Rural East New Britain (18) Gazelle 1 03.Lassul Baining 14.Yalam 022.Yalam 105 Rural East New Britain (18) Gazelle 1 04.Livuan/Reimber 20.Ramalmal 001.Ramalmal 153 Rural 104 Appendices Table 35 Selected census units (outside of NCD) (continued) # of District Local Level HOUSE- Province District # Government Ward Census Unit HOLDS East New Britain (18) Gazelle 1 05.Vunadidir/Toma 16.Bitakapuk No.2 013.Bitakapuk No.3 79 Rural East New Britain (18) Pomio 3 10.Central/Inland 19.Mukulu 003.Mukulu 31 Pomio East New Britain (18) Pomio 3 11.East Pomio 10.Bain 019.Bain Village 75 Rural East New Britain (18) Pomio 3 13.Sinivit Rural 01.Rieit 004.New Camp 61 East New Britain (18) Pomio 3 13.Sinivit Rural 17.Ivon/Gore 552.Ivon Settlement 161 East New Britain (18) Pomio 3 14.West Pomio/ 28.Yauyau 030.Yauyau 34 Mamusi Rural Note: Census Units with highlighted counts may require segmenting (for large units) and combining for small units. The selected Census Units include one small CU, Kariaritsi in Goilala District, for which the preliminary Census count suggests there will be insufficient households to allow a quota of ten to be surveyed. It is therefore recommended to create a pseudo-CU by joining Kariaritsi with its neighbour in the same ward, Kataipa. The merged CU should enable a quota of ten households to be surveyed. Conversely there are eight CUs that appear to be so large that listing and surveying may be logistically burdensome if the entire CU is attempted to be covered. These large CUs are candidates for being segmented, where a sub-unit of the CU is selected at random (e.g., using a sequence of coin tosses, if it is one-half, one-quarter or one-eighth of the CU that is to be selected). The listing and household level selection then occurs only within the segment of the CU, and the count of households in the segment is inflated by the inverse of the sampling fraction to estimate the total size of the CU (e.g., if the segment is one-quarter, then four times the count of households in that segment is used as the estimate for the size of the overall CU). It is important that the decision to segment a CU is made in advance and not by the team on the ground, otherwise listing efforts are likely to be compromised by shirking teams deciding to only partially list a CU, biasing the sample towards the most easily accessible households. The survey is likely to face localized disruptions that either prevent interview teams from reaching a Census Unit, or from fielding the survey once there. A set of 19 replacement CUs has been selected (one per selected district) using a PPeS draw and these are listed in Table 36. It is important that survey funders and implementers determine in advance what level of decision-making unit, from team supervisors up through the overall survey organizer, is allowed to make the decision to abandon attempts in one CU and use the replacement CU from the same district. If control over this decision is not determined in advance the sample integrity will be degraded, as it is likely that teams give up on difficult-to-reach CUs, violating the requirements of a representative sample. 105 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Table 36 Substitute Census Units in Case of Inaccesibility of a Selected CU (one per District) # of District Local Level House- Province District # Government Ward Census Unit holds Central (3) Goilala 2 06.Woitape Rural 08.Aduai 001.Aduai 78 Central (3) Kairuku/Hiri 3 10.Mekeo Kuni Rural 16.Upper Kuni 022.Ifonobu 19 Western Highlands (9) Dei 2 14.Kotna Rural 20.Timbi 055.Timbi 76 Western Highlands (9) Mt Hagen 3 04.Mt Hagen Rural 31.Baisu 010.Tibi 37 Jiwaka (22) Anglimp/South 1 02.South Waghi Rural 26.Pukamil 024.Pukamil 2 65 Waghi Jiwaka (22) Jimi 4 06.Jimi Rural 23.Tsenga 014.Tsenga 84 Eastern Highlands (11) Asaro/Watabung 1 12.Lower Asaro Rural 06.Gamiyuho 063.Noliligeto 18 (Daulo) Eastern Highlands (11) Goroka 5 14.Mimanalo Rural 04.Kabiufa No. 2 403.Gizameka 69 Eastern Highlands (11) Okapa 7 10.East Okapa Rural 05.Yagareba 020.Kume 105 Morobe (12) Bulolo 1 29.Buang Rural 11.Mapos 1 014.Mapos 1 98 Morobe (12) Finschafen 2 06.Hube Rural 29.Qwakugu 009.Qafin 39 Morobe (12) Huon 3 11.Wampar Rural 19.Chivasing 004.Chivasing 134 Morobe (12) Lae 5 16.Lae Urban 82.Lae City 065.Sumiho Street 150 Morobe (12) Markham 6 19.Wantoat/Leron Rural 16.Gumia 008.Kawan 23 Morobe (12) Menyamya 7 21.Kome Rural 14.Hartingli 004.Lagai 188 Madang (13) Rai Coast 4 12.Rai Coast Rural 09.Kakimar 040.Kalaleng 36 Madang (13) Sumkar 5 13.Karkar Rural 12.Kaul 1 009.Kaul 2 122 East New Britain (18) Gazelle 1 05.Vunadidir/Toma Rural 10.Vunakabi 010.Vunakabi 126 East New Britain (18) Pomio 3 12.Melkoi Rural 17.Lausus 026.Lausus 78 In the NCD there are fewer logistical barriers to reaching selected CUs, but the rate of refusal or non-contact of households is likely to be much higher. It is not clear therefore that replacement CUs are needed for the NCD, so only the target of 25 selected CUs are listed (in Table 37). These have been selected using PPeS, where the implicit stratification is according to Census Division (Ward) numbering, starting from Gerehu, proceeding through Waigani, Gordons, Boroko, Town, and out to Bomana. This is broadly a North to South and then a West to East direction; with the selected CUs including urban villages, settlements, and employer-based housing schemes. Several of the selected CUs are quite large and will require segmenting prior to the selection of households, and this may be aided with the Census maps and also the aerial images from Google Maps which have a high resolution for NCD. 106 Appendices Table 37 Selected census units in NCD # of Ward (Census Division) Census Unit Households 80.Gerehu Urban 005.Karukas 245/1 97 80.Gerehu Urban 027.Hariva Street 300 99 80.Gerehu Urban 047.Agolo Dr. 242/238 83 81.Waigani/University 007.Heduru Pl. 50 81.Waigani/University 028.Gull St. 39 & 40 84 82.Tokarara/Hohola Urban 001.Helai Avenue 576 82.Tokarara/Hohola Urban 015.Iduhu St. 92 82.Tokarara/Hohola Urban 037.Taraga Rd. 109 82.Tokarara/Hohola Urban 052.Murray Barracks 704 83.Gordons/Saraga Urban 013.Henao Dr. 73 170 83.Gordons/Saraga Urban 038.Air Niugini Village 290 83.Gordons/Saraga Urban 046.Dunlin Cres. 97 103 83.Gordons/Saraga Urban 059.6 Mile Dump S’mnt 121 84.Boroko/Korobosea Urban 037.Korobosea Dr. 57 84.Boroko/Korobosea Urban 074.Taurama Barracks 470 84.Boroko/Korobosea Urban 081.Pari Village 595 85.Kilakila/Kaugere Urban 015.Kila RPC Police Barracks 102 85.Kilakila/Kaugere Urban 032.Pruth St. 142 85.Kilakila/Kaugere Urban 057.Savalea/Bundi S’mnt 176 86.Town/Hanuabada Urban 022.Brampton St. 85 86.Town/Hanuabada Urban 036.Elevala,Lahara,Gabi 562 86.Town/Hanuabada Urban 058.Red Sea area/yard 220 87.Laloki/Napanapa Urban 512.8-Mile S’mnt 601 88.Bomana Urban 407.Popondetta S’mnt (ATS) 421 88.Bomana Urban 8508.NDST Company 35 Note: Bold values are Census Units that are likely to require segmenting. Stage Four Selection of Households (and Listing) Listing of all households in the selected CU is required for two purposes: for ensuring that all parts of the CU are represented amongst the surveyed households, and for checking on the size of the CU that affected its selection probabilities (with the actual size from the listing used to reweight compared with the estimated size used for the initial selection). The team walks around in every part of the CU, accompanied by a 107 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea guide who is a member of the community. If possible, find a person who worked with the Census in this community or someone with similar knowledge of the community and ask them to be your guide. Make sure you go to all parts of the CU, including outlying hamlets. In hamlets, or in any place far from the centre, always check: “Do these people belong to (Name) village or Census Unit” In every part of the village, ask the guide about every house: “Who lives in this house? What is the name of the household head?” Note that you do not have to visit every household. At best, you just need to see each house but you do not need to go inside it or talk to anyone who lives there. Even the rule of seeing each house may be relaxed if there are far away households for which good information can be provided by the guide. The most important thing in the listing is to be sure that you list all the households and only the households belonging to the named village or Census Unit (or subset of the Census Unit if it is to be segmented). It does not matter in what order you list the households as long as they are all listed. If the Census Unit is large, such that it exceeds approximately 200 households, or is very dispersed (as in parts of the Highlands) it is not practical to attempt to list the whole Census Unit, so a decision is made in advance to split the Census Unit into smaller areas (perhaps groupings of clans, or based on a geographical feature). First, a local informant should communicate the boundaries of the Census Unit and any natural or administrative sub-units within the larger Census Unit (such as hamlets or canyons/valleys). The sub-units should be big enough to allow for the selection of a set of households (about 40 or more), Once the sub-unit is defined, its boundaries should be clearly described. Then one of the smaller units is randomly selected and the procedures outlined above are then followed to complete the listing and interviewing just in that sub-unit. After the listing is complete, check that all lines on the listing form are numbered consecutively with no gaps, from start to finish. The number on the last line should be exactly the number of households listed. If the list is long (say more than 50 households) interviewers may encounter difficulties when looking for their selected household. One useful way to avoid this is to show approximately the place in the list where certain landmarks are. This can be done by writing in the margin, CHURCH or STORE or whatever. The sampling within the CU uses circular systematic selection, which differs from the selection procedure described above for provinces, districts and CUs. Let M be the total number of households listed, and the sampling interval L is calculated as (M÷10), rounding to the nearest whole number. Let R be a random number with 3-digit decimals between 0.000 and 0.999 (if the proposal to use tablets for interviewing is implemented, the random number can be calculated in the field, otherwise the listing forms may need to have random numbers pre-written on them). Multiply M by R and round to the nearest whole number, which gives the 1st selection. Note that this could 108 Appendices occur anywhere on the list since the random number lies in the 0–1 interval. Enter SEL against this line in the selection column of the list. Count down the list, beginning after the 1st selection, a distance of L lines to get the 2nd selection, then another L to get the 3rd, etc. When you come to the bottom of the list, jump back to the top as if the list were circular. Stop after the 10th selection. There is now a systematic sample of the households in the Census Unit (or segment of the Census Unit). In order to allow for replacement households in case of refusals, advance down L lines from the 10th selection, and that household is a Reserve (if this happens to be an already selected household, go down one further line). Additional reserves can be found by moving down L lines each time, as needed. It is important that the Listing Form be kept as a record of the selection, and that records be kept when replacement households are used. These details also are needed because the count of households in the CU is important for calculating the final sampling weights. Weights and Variance Calculations The overall sampling weight attached to each household is the product of three intermediate weights: the first stage (province) weight, the second stage (district) weight, and the third stage (CU) weight adjusted for the selection of a particular number of households within the selected Census Unit. Consequently, the final survey weights can only be calculated once the survey fieldwork is complete, since the achieved sample size may differ from the target sample size in some Census Units (and this is only known ex post), and also because the estimate of the number of households in the CU from the survey listing is needed in the calculation of the final weights. However, the structure of the required calculations for each of the weights can be outlined in advance, as follows: ■■ First stage weight, which represents the inverse of the first-stage selection probability assigned to each province. This Province Weight (PW) depends on five values, where only three of these were known at the time of selection: the number of provinces to be selected (7), the preliminary count of households in all provinces excluding NCD, which is denoted M (=1,367,029), th and the preliminary count of households in the i province, denoted mi. The remaining two values become available once the final Census counts are released, and are the ratios of the final count of households, M' to the preliminary count of households, M in all provinces outside the NCD (R = M'/M) and this same ratio of the final count to the preliminary count of th households in the i province (ri = mi'/mi). m m PWi = 1 M R = 1 M M ' i = 1 M M ' i = 1 M ' 7 mi ri 7 mi M mi ' 7 mi M mi ' 7 mi ' 109 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea The weight calculation above shows how various terms cancel once the ratios of final counts to preliminary counts are introduced. The cancelation of terms highlights the fact that there is no presumption that the measures of size used in the initial selection, M and mi are fully accurate. Hence, ongoing debates in PNG about the reliability or timeliness of the 2010 Census are not germane to the sample integrity for the Financial Capability Survey. If the initial counts are accurate, the sampling (random) error is likely to be smaller, which is an advantage, but irrespective of whether they are accurate or not, there will be no bias, provided the above weights are used. Specifically, in the set of fractions after the second equals sign, the value mi in the numerator of the final term and the value M in the denominator of the second term are introduced in order to cancel the varying probabilities which were initially used in selecting the provinces. This cancelling will only occur if we use identically the same values for M and mi in the weighting formula as those used in selecting the sample. That the preliminary counts used at that stage may have been inaccurate does not matter because they are going to be updated by introducing the values M' and mi', coming from the finally released counts. ■■ Second stage weight, representing the inverse of the second-stage selection probability assigned to each district. This District Weight (DW) depends on five values that are comparable to those used for calculating the Province Weight. The number of districts to be selected (19), the preliminary count of households in the seven selected provinces, which is denoted N (=613,652), th and the preliminary count of households in the i district, denoted ni. The remaining two values become available once the final Census counts are released, and are the ratios of the final count of households, N' to the preliminary count of households, N in the seven selected provinces (K = N'/N) and this same ratio of the final count to the preliminary count of households th in the i district (ki = ni'/ni ). DWi = 1 N K 19 ni k i ■■ Third stage weight, representing the product of the inverse of the third-stage selection probability assigned to each Census Unit and the inverse of the selection probabilities for each household within Census Unit j. Within each of th the i selected districts, the selection probability for the j Census Unit (CWij ) depends on five values that are comparable to those used for calculating the District Weight and the Province Weight (except they now vary district-by- district). The first three of these values are known in advance of the survey: the number of Census Units to be selected in each district (5), the preliminary count of households in the selected district, which is denoted Di, and the th preliminary count of households in the j Census Unit, denoted cj. The 110 Appendices remaining two values needed for calculating the selection probability for each Census Unit in each district become available once the final Census counts are released and once the survey fieldwork is complete. These are the ratios of the final count of households, Di' to the preliminary count of households, Di in each selected district (Qi = Di'/Di) and this same ratio of the final estimate of the number of households in the Census Unit (from the listing done for th the survey) to the preliminary Census count of households in the j Census Unit that the selection was based on (qj = cj'/cj). The remaining component of this third stage weight is the inverse of the selection probabilities for each household within Census Unit j, given by the ratio of the final estimate of the number of households in the Census Unit (from the listing done for the survey), which is denoted by cj' to the actual number of households with completed surveys, h (which is targeted to be 10). D Q c′j CWij = 1 i i × 5 cj qj h The final weight will be the product of the CWij , the DWi , and the PWi , and is interpreted as the number of households nationally that are represented by each household that is surveyed. The use of the correct weights will ensure that estimates of sample means, proportions and totals are nationally representative. However, there also is interest in measures of uncertainty (standard errors and variances) due to the point estimates coming from a sample rather than from a complete enumeration of all households in PNG. These measures of sampling error have to allow for the complex nature of the sample, in terms of the weights, the stratification, and most especially the clustering resulting from the multi-stage selection process. In general, clustered samples have less variability in them than simple random samples (SRS) of the same size, since observations from the same cluster are more alike than observations drawn at random. With less variability, the sampling errors are wider than they would be for a simple random sample of the same size and this is typically shown in terms of a “design effect” (the ratio of the actual variance to what the variance would be with a sample of the same size but drawn using SRS). The design effect for multi-stage clustered samples, like the one drawn here, is typically larger than if the sample of selected Census Units had been drawn in a single stage. Of course, offsetting the higher sampling errors for a multi-stage clustered design compared with, say, a two-stage design are the potentially smaller non- sampling errors that result from being able to better control survey fieldwork that is operating only in a limited number of provinces and districts rather than being thinly spread over the whole country. Moreover, the sampling design effects can be calculated but the effects of non-sampling errors remain unknown and introduce inaccuracies and non-comparable survey conduct in ways that are impossible to know. 111 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea What is important at the analysis stage when calculating the sampling errors is to use software that can correctly account for the particular sampling design used. One option, which may be familiar to some users of the final survey data, would be SPSS. However, within the Complex Samples module of SPSS 17, the Sampling and Analysis Preparation wizards allows a maximum of only three stages in a sample design whereas four stages are used here. A better option is probably Strata, which is more flexible and allows more stages. The specific information that is needed for the command that would be used to declare the nature of the survey data is as follows: ■■ The final sampling weight (the product of the CWij, the DWi, and the PWi), hhwght ■■ A categorical variable to indicate each province (the primary sampling unit), prov ■■ A categorical variable to indicate each district (the secondary sampling unit), district ■■ A categorical variable to indicate each Census Unit (tertiary sampling unit), CU ■■ A categorical variable to show the first level stratification between NCD and the rest of PNG, and the female literacy-based stratification, strata1 ■■ A categorical variable to show which category of female literacy and disadvantage each of the selected districts came from, strata2 ■■ Finite population corrections, which indicate that the sampling was without replacement and that the population sampled from was small enough for that to matter. These would be the total number of provinces, nprov; the number of districts available to be selected, ndist; and the number of Census Units in each selected district that were available to be selected, ncu The appropriate command to issue in Strata to declare the nature of the sample design then would be: svyset prov [pw=hhwght], strata(strata1) fpc(nprov) singleunit(scaled) || district, strata(strata2) fpc(ndist) || cu, fpc(ncu) The || indicate each subsequent stage of the sampling, and the “single unit (scaled)” command tells Strata how to handle stratum with only one sampling unit when calculating variances for the survey data. Once Strata is informed of the sampling design, this information is preserved in the dataset, and users of the data have a choice over several different approaches to calculating the variance of any statistics of interest, including Taylor series linearization, jacknife, balanced and repeated replication, or bootstrapping. 112 Appendices Appendix 3b: Sampling Weights for the PNG Financial Capability Survey John Gibson. 21 April 2015 Background The original sample selection was based on a stratified four-stage random sample design, with preliminary counts from the 2011 Census of Population providing the sampling frame. Due to the limited survey budget and the logistical difficulties of surveying in PNG, the first stage selection drew a stratified sample of just seven provinces to ensure that fieldwork was not spread too thinly over the whole country. These provinces included three from the Highlands (Western Highlands, Jiwaka, Eastern Highlands), two from Momase (Madang and Morobe), and one each from the New Guinea Islands (East New Britain) and Papuan (Central) regions. These seven provinces had been selected using female literacy and predicted poverty as the stratifying variables, since it was expected that these variables are likely to be closely related to financial access and financial capability (especially given the focus of the survey on gender differences) so stratifying by these variables would ensure that the sample covered the range of conditions found in PNG. In addition, the original sample had the National Capital District (NCD) included as a separate survey strata. At the second stage, a stratified sample of 19 districts was chosen from the seven provinces, with selection probabilities proportional to estimated size (PPeS, where ‘size’ is the number of households in each district). Six of those 19 districts were from Morobe province, where this high number from one province occurred by chance but also reflects two features of Morobe: it has a wide range of conditions, which means that it is represented in most of the cells of the stratification table, since it has some districts that are amongst the most disadvantaged in PNG while other districts are some of the most well off, in terms of having high rates of literacy and low rates of poverty. Secondly, it has the largest population of any province and so probability proportional to size selection will tend to select areas from this province. The neighbouring province of Madang also had two districts selected. As it turned out, the survey was only able to be fielded in Morobe and Madang. The calculated sample weights therefore need to be revised to reflect this changed scope so that the estimates from the survey may be in some sense representative of conditions in Morobe and Madang. The purpose of this note is to explain the approach taken to re-weighting. It should be read in conjunction with the original sampling report, which also provided details on the appropriate variance estimators to use with this complex sample design. It should be noted that the change in the scope and sample size of the survey should not affect the recommendations made in the original report for the most appropriate 113 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea variance estimator to use when calculating standard errors or testing hypotheses. Nor does the change in focus reduce the representativeness of the results for each of the selected districts. However, it is the case that some reweighting at Census Unit level is also needed (and is discussed below) because the achieved sample sizes within the 44 selected districts were not always the same as the target sample size. In some cases an entire Census Unit was inaccessible and a replacement CU was surveyed, based on a pre-set list of one CU per selected district which had been selected using a PPeS draw. Reweighting of Districts The eight districts from the original selection that were surveyed represent themselves and a certain number of other districts, to add up to the total number of districts in the sample frame. If these districts had been chosen using a random selection (with probability proportional to size), where the sample frame was just the 15 districts in Morobe and Madang (rather than a sample frame of all districts in the originally selected seven provinces) then the district-level weight (DWi) would be straightforward to calculate from three values: a) The total number of households in Morobe and Madang from the 2011 Census counts, which is denoted M (this number equals 227,409) th b) The number of households in the i selected district, denoted mi (this count ranges from 11,120 households to 23,661 households) c) The target number of districts (n=8) DWi = 1 M (1) 8 mi This weight ranges from a low of 1.20 for a large district like Lae to 2.56 for a small district like Finschafen. The intuition for this variation is that Lae has a high probability of being selected, given its large size, while Finschafen has a much lower probability. Thus, Finschafen represents more districts, since it is an example of the small, unlikely-to-be-selected types of districts while Lae does not represent as many other districts since large districts are already more likely to be selected. If the resulting weights are applied, then in the hypothetical situation where all households in each 44  The third and fourth stage selection were that within each district, five Census Units (CUs) were selected with probability proportional to estimated size (PPeS) and within each CU a total of ten households were to be selected, using circular systematic sampling. In addition, a ‘replacement’ CU per district was selected, again using PPeS, in cases where one of the originally selected CUs was inaccessible. 114 Appendices selected district were surveyed, the weighted number of households would equal the 45 total of 227,409 in the 15 districts in Morobe and Madang. The problem with this calculation is that these eight districts were not selected from a sample frame made up of just Morobe and Madang Districts and instead were selected from a much larger frame that was subject to implicit stratification. Therefore the districts that were selected from Morobe and Madang were designed to represent conditions in all of PNG (when combined with the districts in the provinces that were not surveyed) and are not necessarily equally as representative of conditions in Morobe and Madang. One approach would be to treat the implicit stratification (which affected the order in which districts were ranked prior to their initial selection) as if it had been explicit stratification. The weighting calculation above would then be carried out separately within each strata so that selected districts were weighted-up so that the sample would represent all households within that strata. Thus if the selection based on national-level considerations caused ‘too few’ or ‘too many’ districts to be selected from with a particular strata, from the point of view of best representing Morobe and Madang, this reweighting could take care of any shortage or surplus of districts in the sample. The problem however is that not all cells in the stratification table are represented amongst the sample of selected districts, as shown in Table 38 below. The cross-tabulation of the two stratifying variables shows two empty cells, in the sense that Morobe and Madang have no districts with the combination of being Table 38  Districts in Morobe and Madang, by Stratification Groups (Selected Districts shown in italics) Predicted Poverty Group 1 = highest 2 = middle 3 = lowest Menyamya 1 (low) Middle Ramu No Districts Usino/Bundi Female literacy group Tewae/Siasi Bogia 2 (med) Kabwum Bulolo Rai Coast Markham Huon Lae 3 (high) No Districts Finschafen Nawae Sumkar Madang 45  The weights at Census Unit level achieve the same effect, with the weighted sum of the number of surveyed households in each district adding up to the total number of households in that district. 115 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea in the highest poverty group and the highest female literacy group, or being in the lowest poverty group and the lowest female literacy group. Of the other seven cells in the table, only five have districts that were part of the sample, with the cells for the highest poverty group and lowest female literacy group and for the middle poverty and middle literacy group not being represented in the sample. In other words, the approach of up-weighting the selected districts to represent strata totals would not provide any way for the survey results to represent the conditions that are likely to exist in Middle Ramu (a very disadvantaged district) and Kabwum (which is middling, in terms of literacy and poverty). In contrast, the selected districts are drawn heavily from the most advantaged cell in the table, with high female literacy and low poverty, and so are likely to represent better financial capability and access than occurs for 46 Morobe and Madang as a whole. Thus the approach of treating the implicit strata as explicit strata and up-weighting the selected districts to get to strata-level total numbers of households will not work. Instead, an adaptation of a non-parametric approach for reweighting an entire distribution, which was developed by DiNardo, Fortin and Lemieux (1996) and is based on creating a counterfactual density, is used. This approach has been previously used in the survey context by Cameron et al. (2010) to re-weight the sample of a survey that was fielded in just two provinces in Cambodia so that it matched the distribution of a larger, nationwide, survey. The method relies on a probability model (logit or probit) being applied to a dataset where the small sample and the large sample are pooled, and the dependent variable is an indicator for whether each observation comes from the larger sample or the smaller sample. Intuitively, the observations from the smaller sample whose characteristics make them more likely to have come from the larger sample are given a higher weight, while those from the smaller sample that have characteristics that make them less like the observations in the larger sample are given a lower weight. When applied in the current context, this method provides a way to adjust the weights for the selected districts so that they represent all parts of Morobe and Madang, including the two cells in Table 38 that have districts but were not covered by the survey. A probit model was used, with the predictor variables being the female literacy rate and dummy variables for whether each district was in the high poverty or medium poverty categories. The observations are districts, which are weighted by their number of households so as to match the PPeS approach to selection that was used in the original sample design. For most of the selected districts, their conditional (on literacy and poverty) probability of coming from a larger sample (in fact, coming from the full 46  The over-representation of more advantaged districts is increased if additional, unplanned, surveys that were carried out in Madang District are used, since that would mean that 4/9 surveyed districts came from the high literacy–low poverty cell, which contains only one-third (5/15) districts in all of Morobe and Madang. 116 Appendices Table 39 Inputs into the Calculation of District-Level Weights for the Selected Districts Predicted Rescaled PPeS Rescaled PPeS District probability Probability Weight Weight Rai Coast 0.763 1.233 2.091 2.553 Menyamya 0.624 1.008 1.685 1.682 Finschafen 0.654 1.056 2.556 2.674 Markham 0.582 0.941 2.055 1.915 Bulolo 0.582 0.941 1.377 1.283 Lae 0.603 0.975 1.201 1.160 Huon 0.596 0.963 1.623 1.547 Sumkar 0.595 0.962 1.674 1.595 Notes: Predicted probability is from a probit model for the selected districts to be part of a larger sample that includes all districts in Morobe and Madang, and is rescaled to have a size-weighted average of 1. The PPeS weight is if selection was from an un-stratified probability proportional to estimated size selection using a frame made up of all districts in Morobe and Madang, and is rescaled using the values the rescaled probabilities. population) that included all 15 districts in Morobe and Madang is around 0.6 (noting that the eight selected districts contain 134,211 households while all 15 districts contain 227,405 households, so the mean probability is just under 0.6). However, for Menyamya, Finschafen, and especially Rai Coast, the predicted probabilities are higher and reflect the fact that the literacy and poverty conditions found in those districts are under-represented in the sample compared with their prevalence amongst all Morobe and Madang Districts. In other words, Menyamya, Finschafen, and Rai Coast are more like the two districts (Middle Ramu and Kabwum) in the two cells of the stratification table that were not covered by the selected sample, so that more weight needs to be put on them so that they also represent those omitted cells. The predicted probabilities are rescaled, by normalizing by the (number of household) weighted mean predicted probability (0.619) and this rescaling factor is then applied to the PPeS weight that is generated from using equation (1). The resulting weights in the last column of Table 39 are the appropriate ones to apply to each selected district to represent the range of conditions and the total number of households in all districts of Morobe and Madang. Reweighting of Census Units The originally calculated weights for each Census Unit (CU) were based on a selection of five Census Units per district to be surveyed, with ten households selected from each CU. The CUs were selected using PPeS, based on the counts of households in the 2011 Census (and the ‘reserve’ CU for each district was also selected using PPeS). In 117 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea principle, these weights could be updated with more recent estimates of the size of each CU (e.g. from the listing operation prior to the selection of the ten households) but these details were not provided in the survey dataset. Therefore, the only variation in the weights from what was originally calculated is for instances where more or fewer CUs than targeted were surveyed, and instances where more or less than the target of ten households per CU were surveyed. The assumption made by the reweighting is that variation from the target number of households is at random, so that the actually surveyed households from the same CU are up-weighted or down-weighted to compensate for the over- or under-target 47 number of interviews achieved. The same assumption is also made between CUs within the same district—if fewer than five Census Units were surveyed they are up- weighted to compensate for the ones that were meant to be surveyed but were not. The assumption of missing at random is less defensible for CUs than for households within Census Units, since factors such as geographic inaccessibility and religious belief are sometimes the reason for why a CU is not surveyed. But it is difficult to model these factors so as to find surveyed CUs that are more like the unsurveyed ones, so that these more alike ones can be up-weighted. So the less information- demanding approach of missing at random is used here. Final Weights The overall sampling weight attached to each household is the product of two intermediate weights: the first stage (district) weight, which is provided in Table 39 above, and the second stage (CU) weight that has been adjusted for the number of surveyed households within the selected Census Unit and of CUs within the district. The CU weight and the household weight are listed in Table 40 below. The household weights range from 499 to 991 and it should be noted that applying these weights to the final sample gives a weighted total of households that matches the Census count of households for Morobe and Madang. If the CU weight is used for the sample of households within a district, it yields weighted estimates representative for that district (albeit with a small sample size of typically 50 households), given the nature of the random PPeS selection of CUs within districts and circular systematic sampling of households within CUs. For households that provided two sample members, the same household weight should be used for each member. Thus, the data for households with a male respondent and female respondent have more impact on calculated summary 47  Note however that the same principle is not applied to weighting individual interviews because there is no expectation that all households have one adult male and one adult female who would each be eligible to complete the survey. Thus any imbalance in the number of male and female interviews is assumed to reflect a gender imbalance present in the population of these two provinces. 118 Appendices Table 40 Census Unit and Household Weights Dist CU HH Dist # District Ward CU CU Name Weight Weight Weight 1201 Bulolo 17 7 007.Mumengtain 1.283 413 530 1201 Bulolo 2 7 007.Hikiawa 1.283 413 530 1201 Bulolo 80 8 008.Forestry Compound 1.283 413 530 1201 Bulolo 4 411 411.Wara Kalap 1.283 413 530 1201 Bulolo 4 1 001.Moniau 1.283 413 530 1202 Finschafen 30 12 012.Genna 2.674 371 991 1202 Finschafen 9 24 024.Lecko 2.674 371 991 1202 Finschafen 1 507 507.Brown Hospital 2.674 371 991 1203 Huon 19 4 004.Chivasing 1.547 350 542 1203 Huon 8 19 019.Salus 1.547 350 542 1203 Huon 8 9 009.Labubutu 1.547 350 542 1203 Huon 13 416 416.Sepik Settlement 1.547 438 678 1203 Huon 18 436 436.Erap C/School 1.547 389 602 1205 Lae 82 640 640.Kamkumung Market 1.160 473 549 1205 Lae 82 674 674.Busurum Settlement 1.160 473 549 1205 Lae 82 10 010.Quail Crest 1.160 526 610 1205 Lae 82 110 110.Bumbu Polic Barracks 1.160 526 610 1205 Lae 82 178 178.Tent City 1.160 430 499 1206 Markham 1 501 501.Umi/Atzera LG Station 1.915 307 589 1206 Markham 9 8 008.Yanuf 1.915 277 530 1206 Markham 21 18 018.Sauruan 1.915 277 530 1206 Markham 4 1 001.Arawek 1.915 277 530 1206 Markham 20 5 005.Wangat Sayang 1.915 277 530 1207 Menyamya 12 11 011.Epike 1.682 562 946 1207 Menyamya 9 14 014.Hazoi 1.682 562 946 1207 Menyamya 7 4 004.Okanaiwa (4) 1.682 562 946 1304 Rai coast 3 5 005.Gur 2.553 378 964 1304 Rai coast 14 19 019.Ato Settlement 2.553 252 643 1304 Rai coast 4 28 028.Pisangana 2.553 227 578 1304 Rai coast 17 29 029.Singor 2.553 227 578 1304 Rai coast 35 1 001.Bok 2.553 252 643 1304 Rai coast 9 40 040.Kalaleng 2.553 227 578 1305 Sumkar 4 14 014.Kinim 1.595 340 542 1305 Sumkar 15 31 031.Kurubek 1.595 340 542 1305 Sumkar 26 20 020.Marup 1 1.595 340 542 1305 Sumkar 3 9 009.Murukanam 1.595 340 542 1305 Sumkar 17 15 015.Matukar 1.595 340 542 119 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea statistics for the full (gender pooled) sample than do the data from a household with just a single respondent. This allows for the fact that the overall population is made up of households of different size and gender structure. Similarly, in any gender- disaggregated statistics, the household sampling weight is applied to each respondent from the same household. Appendix 4: Field Work Instrument development, translation and piloting were undertaken as part of the development of the World Bank global Financial Capability survey instruments during 2011–2012. Preparations for the field work commencement third quarter 2013. Initially it was intended data collection for the national study would be undertaken in several regional waves, with teams of enumerators recruited for the Momase, Highlands, and Central waves of data collection. The National Statistics Office agreed to assist with enrolment and (initial) local logistics. A MOA was signed between BPNG and NSO. During the period September–December 2013 the governance committees were established and the implementing agency (INA) was selected and the survey instrument and translation was reviewed. The project was launched formally. During the period November 2013–February 2014 logistics were finalised and communication commenced with provincial and district authorities, and provincial NSO offices. Community briefings were held for the initial provinces. Madang and Morobe, were selected to be the first two provinces to be surveyed, with the training for enumerators to be undertaken in Lae, the provincial capital of Morobe. Lae is Papua New Guinea’s second city, and both its industrial capital and the country’s largest port (and the largest port in the South Pacific region, outside Australia and New Zealand). Lae was selected for the training because it is the most central and accessible city for much of the country, with extensive road access to Madang and the Highlands provinces, as well as across much of the large province of Morobe. The initial pilot survey work was undertaken during the training period in Morobe province, and, with the selected enumerators largely from those two provinces, it was logical to proceed with surveying these two provinces first. Moreover, the two provinces are in some ways a microcosm of many of the diverse geographical and social characteristics of the whole of Papua New Guinea. PNG is geographically, economically and ethnically extremely diverse, from its communities in remote valleys in the Highlands provinces, to coastal and lowland communities on some of the country’s islands, coasts and accessible and inaccessible major valleys. It is impossible for any province, or small selection of provinces and districts to represent the entire country’s range of physical and human characteristics, nevertheless, Morobe and Madang do reflect many aspects of that diversity. 120 Appendices Data collection was undertaken during the period February–November 2014. An initial round of data collection was undertaken February–May 2014. This was followed by a review by the Steering Committee. The decision was taken to contract the scope of the initial round of data collection to Morobe and Madang. Data collection resumed July 2014. Two teams on enumerators undertook interviews, initially in Madang and subsequently in Morobe. Tablet based data collection was used for the financial capability survey. Data was held on the tablet and uploaded periodically. This modality proved successful and enabled data to be collected in remote locations, with data upload when proximate to a cell- tower. Enrolment was paper based, as was the location survey. Following the completion of field work data entry was undertaken in Port Moresby by INA. Data entry was completed and checked January 2015. Data entry was not required for the financial capability survey. Appendix 5: Glossary of Financial Terms in Tok Pisin The glossary of financial terms was developed by the PNG Institute for National Affairs to facilitate translation of the Financial Capability survey document from English to Tok Pisin. The translation was reviewed by the expert reference group who also reviewed the survey translation. English Tok Pisin Translation 1. Question Askim/kwesten 2. Income Moni yu kisim long wok bisnis, salim gaden kaikai, o wok moni 3. Self-employment Ol wok yu yet i kirapim long en long kisim moni 4. Business finance Moni bilong bisnis 5. Personal finance Moni bilong yu yet 6. Household finance Moni bilong hauslain 7. Household Hauslain 8. Responsible wholly or partially Em wok we yu yet i save go pas na wokim long en o yu wantaim ol narapela lain long haus 9. Planning Mekim plen 10. Financial decisions Disisen long sait bilong moni 11. Personal spending Moni we yu spendim/usim long yu yet 12. Yes Yes 13. No Nogat 121 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea English Tok Pisin Translation 14. Managing Lukautim 15. Money Moni 16. Regularly Planti taim 17. Sometimes Sampela taim 18. Rarely Wan wan taim tasol 19. Never Nogat wanpela taim 20. Always Olgeta taim 21. Know exactly Save stret 22. Only a rough idea Save liklik tasol 23. Money left over money Leptova moni/moni I stap bihain 24. Necessary items Ol samting yu/hauslain nidim long en 25. Fluctuations in income Moni namba/mak i save go antap na kam daun 26. Major expenditure Bikpela moni bai yu spendim/yusim long baim wanpela bikpela samting 27. Planned future purchase Samting we yu plen pinis long baim long bihain taim 28. Business Bisnis 29. Assets Aset 30. Essential Items Ol samting yu nidim long en 31. Non-essential items Ol samting yu no nidim long en 32. Lend Givim dinau moni 33. Repay debts Bekim dinau moni 34. Borrow money Dinau long moni 35. Run short Ron sot 36. Insufficient/low income Ino inap moni 37. Business losses Bisnis i lusim moni 38. Unexpected expenses Ekspens we yu no save bai kamap na ol i kamap 39. Financial help Halivim long sait bilong moni 40. Overspending Spendim/yusim moa moni long moni yu gat long en 41. Failure to plan ahead Fail long wokim plen bilong bihain taim 42. Day-to-day spending Moni yu spendim/yusim long wanwan dei 43. Afford to borrow Yu inap long kisim na bekim dinau moni 44. Agree Wanbel 45 Disagree No wanbel 46. Agree Strongly Wanbel stret 47. Agree to some extent Wanbel liklik tasol 122 Appendices English Tok Pisin Translation 48. Disagree strongly No wanbel stret 49. Disagree to some extent No wanbel liklik tasol 50. Jointly Yu wantaim narapela 51. Households necessary items Ol samting hauslain nidim long en 52. Describes you Makim/diskraibim yu 53. Affording a loan Kamap long mak bilong kisim dinau 54. Households future expenses Ol samting we hauslain bai baim long bihain taim. 55. Households average monthly Moni namba we hauslain bilong yu i save kisim planti taim insait long wan wan mun. No income ken kauntim ol mun we hauslain i save kisim liklik moni stret na ol mun we hauslain i save kisim bikpela moni stret. 56. Unexpected major expense Bikpela ekspens we yu no save bai kamap long en 57. Expenses Ekspens/ol samting bai yu baim long en 58. Expected expenses Ol samting we yu save bai yu baim long en 59. Emergencies or unexpected Ol samting we yu no save bai kamap we bai yu mas baim long en expenses 60. Household expenses Ekspens bilong hauslain 61. Remittances Moni ol famili salim kam long yu long lukautim yu 62. In full Olgeta 63. Not worried at all No wari 64. A bit worried Wari liklik tasol 65. Very worried Wari stret/tru 66. Old age Taim yu lapun pinis 67. Strategies Ol rot bilong mekim ol wok i kamap na karim kaikai 68. Savings Seivins/moni yu seivim long benk 69. Pension Pensin/ 70. Financial Assets Fainensel Aset/ol samting olsem moni yu gat long ol bank akaunt. 71. Non-financial assets Ol samting olsem haus, kar, na ol narapela samting yu gat long em 72. Insurance Insurens 73. Future Bihain taim 74. Children Pikinini 75. Month Mun 76. Day Dei 77. Year Yia/kristmas 78. Financial products and services fainensel prodak na sevis (ol samting na sevis long sait bilong mekim, lukautim, dinau, na seivim moni) 123 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea English Tok Pisin Translation 79. Terms and conditions Ol samting/lo bai yu mas bihainim na hevi bilong brukim ol lo makim dispela finensel prodak o sevis. 80. Past 5 years Faiv pela yia/kristmas igo pinis 81. Choose makim 82. Information Infomesin/tok save 83. In-kind payments Baim ol samting wantim ol narapela samting na ino long moni 84. Advantages Gutpela bilong en 85. Disadvantages Nogut bilong en 86. Personally Mi yet 87. Short term Taim i makim nau igo inap long wanpela kristmas tasol 88. Thought Tingting 89. Impulsive kirap nogut na mekim ol samting 90. Aspirations samting insait long tingting bilong mi, mi laik mekim long en istap 91. Status sindaun bilong mi 92. Opportunities Ol wei bilong mekim sindaun bilong mi i kamap gutpela moa 93. Undesirable habits Ol pasin nogut bai bagarapin yu 94. Less fortunate Ol turangu lain 95. Returns on investment Interest moni long invesmen 96. Income from interest on savings Interest moni long seivins 97. Income vary from season to season Moni yu save kisim i save senis long taim bilong rain na taim bilong san 98. Private sector Praivet secta/ol bisnis lain 99. Subletting land or housing Putim graun na manmeri peim moni long yusim/putim haus long rent 100. Income steady Moni namba yu kisim long wok bisnis, o salim gaden kaikai o wok moni i save stap wan kain tasol 101. Income varies Moni namba yu kisim long wok bisnis, o salim gaden kaikai o wok moni i save senis senis 102. Help Halivim 103. Day-to-day Wanwan dei 104. Week-to-week Wanwan wik 105. Month-to-moth Wanwan mun 106. Better off now Gutpela moa nau 107. Worse off now Igo nogut moa nau 108. Will be better off Bai kamap gutpela moa 109. Will be worse off Bai igo no gut moa 110. Just about the same Stap wankain tasol 124 Appendices English Tok Pisin Translation 111. Do not know No save 112. Important Impotent/namba wan samting 113. Financial decision Desisin long sait bilong moni 114. Budget Moni plen 115. Save Sevim 116. Invest Inves 117. Loan Dinau moni 118. Borrow money Dinau long moni 119.B Financial advice Fainensel advais/advais long sait bilong moni 120. Spouse Man o meri bilong yu 121. Family Femili 122. Use Credit Usim Dinau Moni 123. Personal Spending Moni we yu yet bai yusim/spendim 124. Borrow money Dinau long moni Appendix 6: Literature Review of Financial Capability Measurement A number of studies have investigated the relationship between financial knowledge and financial behaviour. Using data from the United States, Lusardi and Tufano (2009) find that individuals who have low measured levels of financial knowledge tend to pay minimum balances on credit cards, incur late fees on cards, and use informal sources of credit. Stango and Zinman (2009) show that people who make mistakes in interest and future value calculations tend to borrow more and save less. Lusardi and Mitchell (2009) illustrate that people with low levels of financial knowledge think less about retirement and that most of them have not planned for retirement at all. A survey of Russian households shows that financial knowledge is significantly and positively related to retirement planning involving private pension funds and schemes (Klapper & Panos, 2011). And in Mexico, Hastings and Tejeda-Ashton (2008) conducted a survey that reveals that less knowledgeable individuals tend to choose mutual pension funds with higher fees. These studies tend to measure financial literacy based on questions that test knowledge of the time value of money (inflation), interest rates, compounding, and risk diversification, although the specific measures used vary from study to study (see also Xu and Zia (2012) for a discussion of different measures of financial knowledge). Most studies do not aim to measure financial capability in addition to financial knowledge, 125 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea and thus there is little existing evidence about the relationship between financial capability and financial behaviour. One caveat with the studies mentioned above is that these results are not necessarily causal. They show a correlation between proxies for financial knowledge and outcomes of interest, but these correlations may simply reflect unobserved characteristics of individuals such as their numeracy, ability, parental background, or other such features. Although some studies try to measure these characteristics and try to account for them in the analysis, some of these features may not be measurable and can thus potentially bias the results. A growing literature tries to address this issue by relying on quasi-experimental or experimental variation in the provision of financial education programs to measure the impact of financial knowledge on financial behaviour. The context of these studies varies widely—for example, in terms of the economic environment and the type of individuals targeted through the financial education programs. Compulsory financial education classes taught in high schools have been the subject of a number of studies. Bernheim, Garrett, and Maki (2001) use exogenous variation in high school financial education mandates across U.S. states to show that students exposed to financial education classes save more as adults. However, Cole and Shastry (2008) cast doubt on these findings, showing that they are not robust to controlling for state-fixed effects and examining effects over time. Shorter-term evidence comes from Bruhn et al. (2013) who conducted a randomized experiment providing financial education in Brazilian public high schools. They find positive effects on financial knowledge, attitudes, and behaviours, and an increase in savings rates. These impacts are small in absolute magnitude: a 3 percentage point increase in knowledge, and a 1 percentage point increase in savings. In Germany, Lührmann, Serra-Garcia, and Winter (2012) find teenagers given financial literacy training show increased interest in and knowledge of financial matters, and save more in a hypothetical task, but they do not measure actual savings. Other studies have focused on providing financial education to working adults, recognizing the differences in households’ financial needs and exposure across developed and developing countries. The literature in developed countries tends to study the impact of financial education on planning for retirement or investment portfolio choices. Duflo and Saez (2011) show that participation in seminars discussing retirement savings leads to an increase in retirement plan participation. In the developing country context, impact evaluations of financial literacy training have studied the unbanked, insurance take-up, and migrants. One of the first papers to examine the impact of financial education in a developing country was by Cole, Sampson, and Zia (2011). The authors implemented a field experiment in Indonesia where they offered randomly selected unbanked households a financial education 126 Appendices course geared toward opening a bank savings account. They find that the financial education course had no effect on the likelihood of opening a bank savings account in the full sample, but it had modest effects for uneducated and financially illiterate households. Cai (2011) used a randomized experiment to show that farmers in rural China are more likely to take up crop insurance and become less price sensitive after attending financial education sessions. Gibson, McKenzie, and Zia (2012); Doi, McKenzie, and Zia (2012); and Seshan and Yang (2012) analyse how providing information and financial education affects the behaviour of migrants and their households. Gibson, McKenzie, and Zia (2012) work with migrants in New Zealand and Australia, and find that financial education increases knowledge about remittance transaction costs but does not lead to changes in the amount of remittances sent or use of the cheapest remittance method. Using a sample of Indonesian migrants, Doi, McKenzie, and Zia (2012) find that impacts on financial knowledge, behaviour, and savings are largest when both the migrants and their families receive financial education. The results show that financial education can have large effects when provided at a teachable moment, but that this impact varies according to who is receiving the training. Seshan and Yang (2012) find that Indian migrants in Qatar increase savings after financial education training, but only if they had low financial knowledge to begin with. Overall, the literature thus finds a positive relationship between financial knowledge and use of formal financial products. Impact evaluations of financial education courses suggest that this relationship is, at least in part, causal. However, these evaluations also highlight that financial education courses often only lead to behaviour change for certain groups of individuals—such as those who had low knowledge to begin with— but not for others. In addition, the measured impacts are often small, and participation rates in financial education courses tend to be low. The small effects and low participation rates suggest that classroom-style workshops may not be the best way of conveying financial education to adults, who may not have the time or motivation to attend such workshops. The literature is now moving toward exploring whether innovative channels for providing financial education can affect behaviour. Ongoing studies in India, Peru, South Africa, and the United States (among others) are testing whether the provision of information via videos, radio, mass media, or video games is effective in improving individuals’ financial decisions (see e.g., Berg and Zia (2013)). While literature has mostly focused on financial knowledge so far, it has also touched on concepts related to specific financial capability. Some of the financial education courses studied through impact evaluations try to teach techniques to improve budgeting and monitoring of expenses. For example, Bruhn et al. (2013) find that a comprehensive financial education program in Brazilian high schools leads to an increase in the percentage of students and parents who make a list of expenses. The program also increased saving rates. Other studies have examined the relationship 127 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea between time preferences and saving behaviour. Brown, Chua, and Camerer (2009) conducted a behavioural laboratory experiment and find that individuals with present- biased preferences have a tendency to overspend. Ashraf, Karlan, and Yin (2006) show that commitment savings accounts can help increase savings for individuals with present-biased preferences. However, more research is needed to investigate the relationship between different components of financial capability and the use of different financial products. The financial capability construct used for this study was developed by Kempson and colleagues at the Personal Finance Research Centre at the University of Bristol. Financial capability has become the dominant construct used to examine financial inclusion, financial knowledge and skill, behaviours and attitudes globally. The construct was initially developed within the context of the UK and was published in the landmark FSA UK Financial Capability Study (Atkinson, McKay, Collard, & Kempson, 2007; Atkinson, McKay, Kempson, & Collard, 2006). The FSA study builds on earlier studies by the Personal Finance Research Centre examining financial exclusion. (Refer for example: Kempson & Whyley (1999); Kempson, Whyley, Caskey & Collard (2000); Kempson, Atkinson & Pilley (2004); and Collard & Kempson (2005).) The instrument was modified and tested at twelve low and middle-income countries in 2010–2013 to incorporate unique characteristics and challenges in these countries. The evaluation of these pilots (World Bank, 2013b, 2013d) shows some common features among the low-income countries in measuring the financial capability and designing financial capability policies and interventions, namely access, poverty, location, informality, education and risks. The evaluation finds that these characteristics are bound to influence the priorities of policy makers with objectives of financial literacy programs and their target groups; the way individuals behave “financially” and how they react to interventions to change their behaviour and how financial capability can be measured. 128 References ADB. (2008). Foundation for the Future: A Private Sector Assessment for Papua New Guinea. Manila: Asian Development Bank. AFI. (2013). Putting Financial Inclusion on the Global Map: The 2013 Maya Declaration Progress Report. Bangkok: Alliance for Financial Inclusion. Anderson, E., Kunjil, N., Ngodup, T. K., & Tongia, G. (2013). Papua New Guinea Financial Service Sector Analysis (draft). Port Moresby: Bank of Papua New Guinea. Ashraf, N., Karlan, D. S., & Yin, W. (2006). Tying Odysseus to the Mast: Evidence from a Commitment Savings Product in the Philippines. Quarterly Journal of Economics 121(2), 635–672. Atkinson, A., McKay, S., Collard, S., & Kempson, E. (2007). Levels of financial capability in the UK. Public Money & Management, 27(1), 29–36. Atkinson, A., McKay, S., Kempson, E., & Collard, S. (2006). Levels of Financial Capability in the UK: Results of a baseline survey. London: Financial Services Authority. Bakani, L. M. ( 2014). The future of credit unions given the global regulatory trends. Paper presented at the Pacific Credit Union Technical Congress, Port Moresby. http://www.bis.org/review/r150112g.pdf Berg, G., & Zia, B. (2013). Financial Literacy through Mainstream Media: Evaluating the Impact of Financial Messages in a South African Soap Opera. Washington, DC.: World Bank. Bernheim, B., Douglas, D. M. G., & Maki, D. M. (2001). Education and Saving: The Long Term Effects of High School Financial Curriculum Mandates. Journal of Public Economics 80(3), 435–465. BPNG (2011). [Bank of Papua New Guinea Strategic Plan 2012–2015]. BPNG. (2013). Papua New Guinea National Financial Inclusion and Financial Literacy Strategy Port Moresby: Bank of Papua New Guinea. BPNG. (2015). Update to GCWG on the Implementation of the National Financial Inclusion and Financial Literacy Strategy 2014–2015. Port Moresby: Bank of Papua New Guinea and Centre for Excellence in Financial Inclusion. Brown, A., Chua, Z., & Camerer, C. (2009). Learning and Visceral Temptation in Dynamic Savings Experiments. Quarterly Journal of Economics, 124(1), 197–231. Bruhn, M., de Souza Leão, L., Legovini, A., Marchetti, R., & Zia, B. (2013). Financial Education and Behavior Formation: Large Scale Experimental Evidence from Brazil. Washington, DC.: World Bank. 129 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Bruhn, M., Ibarra, G. L., & McKenzie, D. (2013). Why Is Voluntary Financial Education So Unpopular? Experimental Evidence from Mexico. Washington, DC.: World Bank. Cai, J. (2011). Social Networks and the Decision to Insure: Evidence from Randomized Experiments in China, from http://blogs.worldbank.org/impactevaluations/. Cole, S. A., Sampson, T., & Zia, B. (2011). Prices or Knowledge? What Drives Demand for Financial Services in Emerging Markets? . The Journal of Finance, 66(6), 1933–1967. Cole, S. A., & Shastry, G. K. (2008). Smart Money: The Effect of Education, Cognitive Ability, and Financial Literacy on Financial Market Participation. Cambridge, MA.: Harvard University. Collard, S., & Kempson, E. (2005). Affordable Credit: The Way Forward. Yok: Joseph Rowntree Foundation. Conroy, J. D. (2000). The Role of Central Banks in Microfinance in PNG. In ADB (Ed.), The Role of Central Banks in Microfinance in Asia and the Pacific: Country Studies. Manila: Asian Development Bank. Demirguc-Kunt, A., Klapper, L., Singer, D., & Van Oudheusden, P. (2015). The Global Findex Database 2014: Measuring Financial Inclusion around the World. World Bank Policy Research Working Paper 7255 Washington: The World Bank. Department of National Planning and Monitoring. (2010). PAPUA NEW GUINEA DEVELOPMENT STRATEGIC PLAN 2010–2030. Port Moresby: Papua New Guinea Department of National Planning and Monitoring. DfCD, & INA. (2011). NATIONAL INFORMAL ECONOMY POLICY (2011–2015). Port Moresby: Department for Community Development & Institute of National Affairs. Doi, Y., McKenzie, D., & Zia, B. (2012). Who You Train Matters: Identifying Complementary Effects of Financial Education on Migrant Households. Washington, DC.: World Bank. Duflo, E., & Saez, E. (2011). The Role of Information and Social Interactions in Retirement Plan Decisions: Evidence from a Randomized Experiment Quarterly Journal of Economics, 118(3), 815–842. Fernando, N. (1992). Informal Finance in Papua New Guinea: An Overview. In D. Adams & a. D. A. Fitchett (Eds.), Informal Finance in Low-Income Countries (pp. 119–132). Boulder: Westview Press. Gibson, J., McKenzie, D., & Zia, B. (2012). The Impact of Financial Literacy Training for Migrants. Washington, DC.: World Bank. GPFI, & CGAP. (ND). Global Standard-Setting Bodies and Financial Inclusion for the Poor: Toward Proportionate Standards and Guidance: The Global Partnership for Financial Inclusion. Hastings, J., & Tejeda-Ashton, L. (2008). Financial Literacy, Information, and Demand Elasticity: Survey and Experimental Evidence from Mexico. Cambridge, MA.: National Bureau of Economic Research. 130 References Hayes, G. (2009). Population ageing in the Pacific: A situation analysis Suva: United Nations Population Fund. HM Treasury. (2007). Financial Inclusion: the way forward. London: Office of Public Sector Information Retrieved from http://www.hm-treasury.gov.uk. Howes, S., Mako, A. A., Swan, A., Walton, G., Webster, T., & Wiltshire, C. (2014). A Lost Decade? Service Delivery and reforms in Papua New Guinea 2001-2012. Port Moresby, PNG: The National Research Institute. IMF, & World Bank. (2011). Financial Sector Assessment Program: Technical Note Financial Inclusion. Washington: International Monetary Fund & The World Bank. Kempson, E., Atkinson, A., & Pilley, O. (2004). Policy Level Responses to Financial Exclusion in Developed Economies: Lessons for developing countries. Bristol: Personal Finance Research Centre, University of Bristol. Kempson, E., & Whyley, C. (1999). Kept Out or Opted Out? Understanding and Combating Financial Exclusion. Bristol: The Policy Press. Kempson, E., Whyley, C., Caskey, J., & Collard, S. (2000). In or Out? Financial Exclusion: A Literature and Research Review. London: Financial Services Authority. Klapper, L., & Panos, G. A. (2011). Financial Literacy and Retirement Planning: The Russian Case. World Bank: Washington, DC. Lührmann, M., Serra-Garcia, M., & Winter, J. (2012). The Effects of Financial Literacy Training: Evidence from a Field Experiment with German High-School Children. Munich: University of Munich. Lusardi, A., & Mitchell, O. S. (2009). How Ordinary Consumers Make Complex Economic Decisions: Financial Literacy and Retirement Readiness Dartmouth University. Lusardi, A., & Tufano, P. (2009). Debt Literacy, Financial Experiences, and Overindebtedness NBER Working Papers. New York: National Bureau of Economic Research. National Strategic Plan Taskforce. (2011). PAPUA NEW GUINEA VISION 2050. Port Moresby: Government of Papua New Guinea. NEFC. (2014). The Thin Blue Line: Technical Report July 2014. The Methodology and Results of the 2011 Cost of Sub-national Services Study (O&M) Port Moresby: National Economic and Fiscal Commission. Seshan, G., & Yang, D. (2012). Transnational Household Finance: A Field Experiment on the Cross-Border Impacts of Financial Education for Migrant Workers, from http://www-personal.umich.edu/~deanyang/papers/seshan%20yang%20 2012%20transnational%20household%20finance.pdf. Sibley, J. E., & Liew, J. (2011). Minimum Adult Financial Competency Framework for Low Income Households in Pacific Island Countries. Suva: Pacific Financial Inclusion Programme. Stango, V., & Zinman, J. (2009). Exponential Growth Bias and Household Finance. Journal of Finance 64(6), 2807–2849. 131 Financial inclusion and financial capability in morobe and madang provinces Papua New Guinea Stuart, G., Noggle, E., & Sibley, J. (2014). The Papua New Guinea Financial Diaries Study. Suva: Pacific Financial Inclusion Programme. World Bank. (2003). World Bank Sector Review Note Washington: World Bank. World Bank. (2012). DRAFT TECHNICAL REPORT: PAPUA NEW GUINEA POVERTY PROFILE. Washington: The World Bank. World Bank. (2013a). Below the Glass Floor: Analytical Review of Expenditure by Provincial Administrations on Rural Health from Health Function Grants & Provincial Internal Revenue. Washington: The World Bank. World Bank. (2013b). Financial Capability in Low and Middle-income Countries: Measurement and Evaluation. Washington, DC: World Bank. World Bank. (2013c). Key Considerations for Provision of Social Pension in PNG: Policy Brief. Washington, DC: The World Bank. World Bank. (2013d). Measuring Financial Capability: A New Instrument and Results from Low and Middle-income countries. Washington, DC: World Bank. World Bank (2013e). Making Sense of Financial Capability Surveys around the World: A Review of Existing Financial Capability and Literacy Measurement Instruments. Washington, DC, World Bank. Xu, L., & Zia, B. (2012). Financial Literacy Around the World. An Overview of the Evidence with Practical Suggestions for the Way Forward. Washington, DC: World Bank. 132 Financial Inclusion and Financial Capability in Morobe and Madang Provinces, Papua New Guinea — An initial report of the Papua New Guinea National Financial Capability Survey “For many Papua New Guineans, the use of nancial services is rather new. This report will provide a detailed picture of how our population manages their funds and accesses the nancial services so we can develop policies, nancial products and services to enhance nancial inclusion and improve nancial capability in PNG.” Mr. Loi M. Bakani, CMG, Governor at the Bank of Papua New Guinea “Understanding the way banks and nancial products work is important for managing money well and saving for the future. The fact that most bank forms in PNG are only written in English, is an unnecessary barrier to saving and managing money. With this in mind, I’m particularly glad to see an English-Tok Pisin glossary of nancial terms come out of this project. I have no doubt this will play an important role in improving nancial literacy in PNG.” Mr. Paul Barker, Managing Director, Institute of National Affairs Bank of Papua New Guinea Institute of National Affairs