Document of The World Bank Report No: ACS22743 Peru PUBLIC EXPENDITURE REVIEW June 16, 2017 Macroeconomics Trade and Investment Global Practice Caribbean Countries Management Unit Latin America and the Caribbean Region This document has a restricted distribution and may be used by recipients only in the performance of their official duties. Its contents may not otherwise be disclosed without World Bank authorization. Government of Peru Fiscal Year January 1 – December 31 Currency Equivalents (Exchange Rate Effective as of February 23, 2017) Currency Unit = Soles (Sol) US$1.00 = 3.25 i ACKNOWLEDGMENTS This report is part of the joint effort of teams from the Ministry of Economy and Finance of Peru (MEF) and the World Bank (WB), in order to suggest consideration of the Government of Peru (GdP) in areas of action that contribute to carry forward the process of fiscal consolidation and improvement in planned expenditure efficiency. Preliminary advances in this report were coordinated and discussed among MEF officials and the WB at various meetings. Thus, on March 28, 2016 a workshop was held in the city of Lima in which the preliminary outline of the chapters was discussed. In addition, a preliminary presentation was made to the top management of the MEF during the Annual Meetings of the World Bank in October 2016. Finally, technical presentations of the preliminary versions of the sectorial chapters were carried out with the presence of the MEF specialists and The WB between December 2016 and January 2017. It should be added that, during the preparation of this report, methodologies and innovative analysis techniques were applied in topics such as procurement, education, health, citizen security, expenditure rigidity and automatic stabilizers, which will be available to GOP officials. The WB team wishes to express its gratitude for the valuable collaboration provided by the MEF under the leadership of Rossana Polastri (Vice-Minister). The counterparts in MEF made important contributions and suggestions on the framework, the methodology, and the interpretation of the findings. The MEF also made databases and reports available to the WB team that greatly facilitated the work for this report. We appreciate the contributions, comments, clarifications and suggestions of the MEF officials, including Rossana Polastri (Vice-Minister), Franklin Huaita (Advisor), Rodolfo Acuña (General Director of Public Budget), José Armando Calderón (Director of Thematic Budget), Rosamaría Dasso (Advisor), José Luis Paco (Director, Budget), Cynthia Flores (Advisor), Jorge Noziglia (Director) and Pedro Casavilca (Advisor). We also appreciate the contributions and feedback from Eva Sifuentes, Rosario Zavaleta, Lucía Sarmiento, Marco Antonio Criollo, Mónica Patricia Sime, Denice Cavero, Lisseth Ramsden, Víctor Bocangel, Nelly Huamaní, Alfonso Gutierrez, José Luis Paco, Reyna Cristina Mendoza, Luz Pilar Soko, Milagros Medina, Gabriela Cuadra, Heidi Rodrich and Henry Espinoza. The Peru's Public Expenditure Review was prepared by a World Bank team led by Marc Schiffbauer (TTL and Senior Economist, GMF04), Pedro Rodriguez (TTL and Program Leader, LC6), Daniel Barco (Economist, GMF04), and Lars Christian Moller (GMF08), who had the strategic orientation of Alberto Rodriguez (Country Director, LC6), Pablo Saavedra (Practice Manager, GMF04), and Doerte Doemeland (Lead Economist, GMF04). The report is based on six articles prepared by World Bank specialists and consultants with international experience in each area. Chapter one was prepared by Daniel Barco (Economist, GMF04); chapter two by Alexandre Borges de Oliveira (Lead Economist, GGO04); chapter three by Caroline van den Berg (Senior Water Specialist, GWA04); chapter four by Christel M. J. Vermeersch (Senior Economist, GHN04); chapter five by Inés Kudo (Senior Education Specialist, GED04) and Hugo Ñopo (Consultant); chapter six by Erik Alda (Consultant) and chapter seven by Kavita Sethi (Senior Transportation Specialist, GTI04). In addition, valuable contributions were made in the preparation of this document Carlos Oliva (Consultant), Rong Qian (Economist, GMF04), Julio Velasco (Research Analyst, GMF04), Heinrich Bofinger (Consultant, GFM04), Melanie Laloum (Consultant, GFM04), Giselle Velasquez (Program Assistant, GMF04), Iris Marmanillo (GWA04), Guillermo Romero (Temporary, LCC6C) and Emilia Abusada (Temporary, GMF04). ii The team is grateful for the collaboration of Luisa M. Yesquén Solari (Program Assistant, LCC6C) and Ingrid Parró (Temporary, LCC6C). iii ABBREVIATIONS AND ACRONYMS AATE Autoridad Autónoma del Sistema Eléctrico de Transporte Masivo de Lima y Callao (Autonomous Authority of the Electric Mass Transport System for Lima and Callao) ABE Alternative Basic Education ADP Adjudicación Directa Pública (Direct public awarding) ADS Adjudicación Directa Selectiva (Selective direct awarding) AIDS Acquired Immune Deficiency Syndrome AMC Adjudicación de menor cuantía (Awarding of minor amounts) ANA Autoridad Nacional del Agua (National Water Authority) APN Autoridad Portuaria Nacional (National Port Authority) AUS Aseguramiento universal de salud (Universal health insurance) BCRP Banco Central de Reserva del Perú (Central Bank of Peru) BEPS Base erosion and profit shifting BID Banco Interamericano de Desarrollo (Inter-American Development Bank) BIE Intercultural Bilingual Education BOOST World Bank database with budgetary variables from official sources CAPEX Capital expenses CAS Contratación Administrativa de Servicios (Administrative services contract) CENACOM Censo Nacional de Comisarías (National Census of Police Stations) CEPAL Comisión Económica para América Latina y El Caribe (Economic Commission for Latin America and the Caribbean) CGR Contraloría General de la República (Office of the General Comptroller) CIT Corporate Income Tax CNSV Comisión Nacional del Servicio Civil (National Civil Service Commission) COAR Colegio de Alto Rendimiento (School for high-performance students) COFIDE Corporación Financiera de Desarrollo (Official Financial Agency for Development) CONASEC Consejo Nacional de Seguridad Ciudadana (National Citizen Security Council) CORPAC Corporación Peruana de Aeropuertos y Aviación Comercial (Peruvian Corporation of Airports and Commercial Aviation) CPM Carrera Pública Magisterial (Public Teaching Career) CREMA Contrato de rehabilitación y mantenimiento (Rehabilitation and maintenance contract) CRFA Centros Rurales de Formación en Alternancia (Rural Alternation Training Centers) DCP Dirección de Crédito Público (Directorate of Public Debt) DEA Data envelopment analysis DGAA Dirección General de Agua y Alcantarillado (General Directorate of Water and Sewerage) DGCT Dirección General de Circulación Terrestre (General Directorate of Land Traffic) DIRESA Dirección Regional de Salud (Regional Health Office) DNI Documento Nacional de Identificación (National identification document) DPP Dirección de Presupuesto Público (Directorate of Public Budget) DRE Dirección Regional de Educación (Regional Educational Directorates) iv ECA Estándares de calidad ambiental (Environmental quality standards) ECE Evaluación Censal de Estudiantes (Census Student Assessment) ENAHO Encuesta Nacional de Hogares (Housholds Survey) ENAPRES Encuesta Nacional de Programas Estratégicos (National Survey of Strategic Programs) ENDES Encuesta Demográfica y de Salud Familiar (Population and Houshold Health Survey) EPS Empresa Pública de Servicios (Water and waswater company) ESCALE Estadística de la Calidad Educativa (Statistics for Quality in Education) EsSalud Company for Health Public Insurance FEF Fondo de Estabilización Fiscal (Fiscal Stabilization Fund) FISSAL Fondo Intangible Solidario de Salud (Intangible Solidarity-based Health Fund) FONAFE Fondo Nacional de Financiamiento de la Actividad Empresarial del Estado (National Fund for the Financing of the State's Business Activity) FONAVI Fondo Nacional de Vivienda (National Housing Fund) FONCODES Fondo de Cooperación para el Desarrollo Social (Cooperation Fund for Social Development) FONCOMUN Fondo de Compensación Municipal (Municipal Compensation Fund) FONCOR Fondo de Compensación Regional (Regional Compensation Fund) FUA Formato Único de Atención (Unique format for attention) GDP Gross Domestic Product GOP Government of Peru GL Gobierno Local (Local Government) GN Gobierno Nacional (National Government) GORE Gobierno Regional (Regional Government) GR Gobierno Regional (Regional Government) HDM-4 Highway Design and Maintenance Standards Model HIV Human Immunodeficiency Virus HWO Health World Organization IAFAS Instituciones administradoras de fondos de aseguramiento en salud (Institutions administering health insurance funds) IBNET International Benchmarking Network for Water and Sanitation Utilities ICT Information and Communication Technologies IDB Inter-American Development Bank IGV Impuesto General a las Ventas (general sales tax) IIEE Instituciones educativas IIRSA Iniciativa para la Integración de la Infraestructura Regional Suramericana (Initiative for the Integration of Regional Infrastructure in South America) IMF International Monetary Fund INCOR Instituto del Corazón (Heart Institute) INEI Instituto Nacional de Estadística e Informática (National Institute of Statistics and Informatics) INFObras Sistema Nacional de Información de Obras Públicas (Public Works Information System) v INVIAS Instituto Nacional de Vías (National Institute of Roads) IPE Instituto Peruano de Economía (Peruvian Institute of Economics) ISC Impuesto Selectivo al Consumo (consumtion tax on selective products) IT Information Technologies JASS Juntas de Administración de Servicios de Saneamiento (Sanitation Services Administration Boards) JEC Jornada Escolar Completa (Full-school day) KfW Kreditanstalt für Wiederaufbau (German Development Bank) KM/Km./km Kilometers KPGM Professional service company KPI Key performance indicators LAC Latin-America and the Caribean LAPOP Latin American Public Opinion Project LMR Lima Metropolitan Region MEF Ministerio de Economía y Finanzas (Ministry of Economics and Finance) MFMod Macro-Fiscal Model of the World Bank MIDIS Ministerio de Desarrollo e Inclusión Social (Ministry of Development and Social Inclusion) MINAM Ministerio del Ambiente (Ministry of Enviromental Affairs) MINEDU Ministerio de Educación (Ministry of Education) MININTER Ministerio del Interior (Ministry of Homeland Security) MINSA Ministerio de Salud (Ministry of Health) MMF Multi-annual macroeconomic framework MMMR Marco Macroeconómico Multianual Revisado (Revised Multi-Annual Macroeconomic Framework) MTC Ministerio de Transportes y Comunicaciones (Ministry of Transport and Communications) MVCS Ministerio de Vivienda, Construcción y Saneamiento (Ministry of Housing, Construction and Sanitation) MDG Millennium Development Goals MUS$ Million dollars of United States of America N/A Not available NFPS Non-financial public sector NGO Non-governmental organization NPV Net Present Value NRW Non-revenue water OCCR Operating cost-coverage ratio OECD Organization for Economic Co-operation and Development OEFA Organismo de Evaluación y Fiscalización Ambiental (Agency for Environmental Assessment and Control) OGPP Oficina General de Planeamiento y Presupuesto (General Unit for Planning and Budget) ONP Oficina de Normalización Previsional (Office of Pension Normalization) vi OPEX Operational expenses OSITRAN Organismo Supervisor de la Inversión en Infraestructura de Transporte Público (Supervisory Organization for Investment in Public Transport Infrastructure) OTASS Organismo Técnico de la Administración de Servicios de Saneamiento (Technical Organization of Sanitation Services Administration) O&M Operation and Maintenance PAAG Programa de Administración de Acuerdos de Gestión (Program for the Administration of Management Agreements) PAHO Pan-American Health Organization PEAS Plan Esencial de Aseguramiento en Salud (Essential Health Insurance Plan) PEN Peruvian Nuevo Sol PER Public Expenditure Review PIA Presupuesto Institucional de Apertura (Institutional Opening Budget) PIM Presupuesto Institucional Modificado (Institutional Modified Budget) PISA Programme for International Student Assessment PMO Project management office PNIE Plan Nacional de Infraestructura Educativa (National Plan for Educational Infrastructure) PNP Policía Nacional del Perú (National Police) PNSR Programa Nacional de Saneamiento Rural (National Rural Sanitation Program) PNSU Programa Nacional de Saneamiento Urbano (National Program for Urban Sanitation) PPP Public-Private Partnership PPP Purchasing Power Parity ProInversión Agencia de Promoción de la Inversión Privada de Perú (Peruvian Agency for the Promotion of Private Investment) PRONOEI Programa No Escolarizado de Educación Inicial (Out-of-school preschool education program) PTAR Planta de tratamiento de aguas residuales (Residual water treatment plant) PVD Provías Descentralizado (Decentralized Pro-Routes) PVN Provías Nacional (National Pro-Routes) RAT Régimen de Apoyo Transitorio (Transitory Support Regime) RBB Results-based budgeting RDR Recursos directamente recaudados (Directly collected resources) RENAMU Registro Nacional de Municipalidades (National Registry of Municipalities) RENAT Registro Nacional de Antecedentes de Tránsito (National Registry of Traffic Records) RENIEC Registro Nacional de Identificación y Estado Civil (national Register of Identity and Civil Status) S.A. Sociedad Anónima (Limited company) SBA Special basic Education SD Standard deviation SEACE Sistema Electrónico de Contrataciones del Estado (Electronic System of State Contracts) vii SEDALIB Servicio de Agua Potable y Alcantarillado de La Libertad, Sociedad Anónima (Potable Water and Sewer Service Company of La Libertad) SEDAPAL Empresa de Servicio de Agua Potable y Alcantarillado de Lima (Potable Water and Sewer Service Company of Lima) SERPOST Servicios Postales del Perú (Peruvian Postal Services) SIAF Sistema Integrado de Administración Financiera (Integrated Financial Management System) SINASEC Sistema Nacional de Seguridad Ciudadana (National Citizen Security System) SINEACE Sistema Nacional de Evaluación, Acreditación y Certificación de la Calidad Educativa (National System of Evaluation, Accreditation and Certification of Educational Quality) SIS Sistema Integrado de Salud (Integral Health System) SISMED Sistema Integrado de Suministro de Medicamentos e Insumos Médico-Quirúrgicos (Integrated System for the Supply of Medicine and Medical-Surgical Inputs) SNIP Sistema Nacional De Inversión Pública (National System for Public Investment) SOSEM Sistema Operativo de Seguimiento y Monitoreo (Monitoring Operating System) SUNASS Superintendencia Nacional de Servicios de Saneamiento (National Superintendence of Sanitation Services) SUNAT Superintendencia Nacional de Aduanas y de Administración Tributaria (National Superintendence for Customs and Tax Administration) SUNEDU Superintendencia Nacional de Educación Superior Universitaria (National Superintendence of University Higher Education) SUSALUD Superintendencia Nacional de Salud (National Superintendence of Health) SUV Sport utility vehicle TERCE Tercer Estudio Regional Comparativo y Explicativo (Third Regional Comparative and Explanatory Study) UE Unidades escolares (education unit) UGEL Unidad de Gestión Educativa Local (Local Educational Management Unit) UMIC Upper middle-income countries UNESCO United Nations Educational, Scientific and Cultural Organization UNODC United Nations Office on Drugs and Crime USAID United States Agency for International Development VAT Value aggregate tax VfM Value for money VMCS Vice Ministerio de Construcción y Saneamiento (Vice-Ministry of Construction and Sanitation) VRAEM Valle de los ríos Apurímac, Ene y Mantaro (Valley of the Apurímac, Ene and Mantaro rivers) WB World Bank WDI World Development Indicators WEF World Economic Forum WHO World Health Organization WSS Water and Sanitation Services viii PERU PUBLIC EXPENDITU REVIEW (P158618) TABLE OF CONTENTS Overview ...................................................................................................................................... 16 Introduction ............................................................................................................................... 16 PART I Cross-cutting Issues..................................................................................................... 18 PART II: Sector Analyses......................................................................................................... 23 Part I – CROSS-CUTTING ISSUES ........................................................................................ 36 Fiscal Policy – Trends and Challenges .............................................................................. 36 Introduction ............................................................................................................................... 36 General Fiscal Dynamics .......................................................................................................... 37 Recent dynamics on the revenue side ....................................................................................... 41 Recent dynamics and selected topics concerning expenditures................................................ 44 Fiscal Sustainability Analysis ................................................................................................... 51 Conclusions ............................................................................................................................... 63 Procurement ......................................................................................................................... 65 Introduction ............................................................................................................................... 65 The scope and framework of public procurement in Peru ........................................................ 65 Strategic sourcing methodology ............................................................................................... 67 Fiscal savings from more efficient procurement strategies ...................................................... 70 Long-term fiscal savings ........................................................................................................... 81 Conclusions ............................................................................................................................... 87 Part II – SECTOR ANALYSES ................................................................................................ 91 Water .................................................................................................................................... 91 Introduction ............................................................................................................................... 91 Spending and Outcomes ........................................................................................................... 91 Spending Efficiency.................................................................................................................. 97 Conclusions and Policy Options ............................................................................................. 109 Annex 1: Institutional Context of Sector ................................... Error! Bookmark not defined. Health .................................................................................................................................. 113 Introduction ............................................................................................................................. 113 Expenditure and Results ......................................................................................................... 113 Peru’s health system: Vision and challenges .......................................................................... 116 Efficiency of the Expenditure on Health ................................................................................ 120 Conclusions ............................................................................................................................. 129 Policies options ....................................................................................................................... 132 Education............................................................................................................................ 137 Introduction ............................................................................................................................. 137 Expenditure and Results ......................................................................................................... 137 Efficiency and Effectiveness of the Expenditure on Education.............................................. 148 Conclusions ............................................................................................................................. 160 Citizen Security .................................................................................................................. 163 Introduction ............................................................................................................................. 163 ix Expenditure and Results ......................................................................................................... 164 Efficiency and Effectiveness of the Expenditure .................................................................... 171 Spending Efficiency of the National Police of Peru (PNP) .................................................... 175 Conclusions ............................................................................................................................. 183 Transport............................................................................................................................ 188 Introduction ............................................................................................................................. 188 Efficiency of Spending ........................................................................................................... 195 Main Challenges and Recommendations ................................................................................ 206 Referencias................................................................................................................................. 208 List of Figures Figure 1-1 Increase in terms of trade and government revenues .................................................. 37 Figure 1-2 Fiscal revenues from mining, other sectors, and totals ............................................... 37 Figure 1-3 Evolution of fiscal revenues and spending, 2004-2016 .............................................. 38 Figure 1-4 Evolution of public spending 2006-2016 .................................................................... 38 Figure 1-5 Public debt and sovereign risk, annual average .......................................................... 40 Figure 1-6 Changes in the fiscal policy stance, 1996-2015 .......................................................... 40 Figure B.1-7 Cyclical balance (proxy of fiscal stabilizers), average of 2005-2014 ..................... 41 Figure 1-8 Declining revenues between 2012 and 2016, by economic cause .............................. 42 Figure 1-9 Declining revenues between 2012 and 2016, by type of tax....................................... 42 Figure 1-10 Tax revenues in Peru and peer countries, 2014......................................................... 43 Figure 1-11 Corporate income tax rate, by country ...................................................................... 44 Figure 1-12 Value-added tax rate, by country .............................................................................. 44 Figure 1-13 Spending increase between 2012 and 2016 by economic classification ................... 45 Figure 1-14 Spending increase between 2012 and 2016 by functional classification .................. 45 Figure 1-15 Main accounts explaining the faster public spending ............................................... 46 Figure 1-16 Types of public spending, % of GDP, 2015.............................................................. 48 Figure 1-17 Spending by sector, % of total spending, 2015 ......................................................... 49 Figure 1-18 Spending rigidity, percentage of total spending ........................................................ 50 Figure 1-19 Spending rigidity, by type of expenditure, percentage of category total .................. 50 Figure 1-20 Fiscal deficit under various scenarios ....................................................................... 53 Figure 1-21 Evolution of the public debt under various scenarios ............................................... 53 Figure 1-22 Adjustment of non-financial expenditure by 2021 under different scenarios ........... 54 Figure 1-23 Simulation of the fiscal impact of the contingencies of the Chavimochic project.... 55 Figure 1-24 IGV Productivity ....................................................................................................... 58 Figure 1-25 Property tax relative to GDP ..................................................................................... 60 Figure 1-26 Public budget and expenditure executed by function, 2015 ..................................... 60 Figure 1-27 Quarterly expenditure, by government level (% of GDP), 2015 .............................. 60 Figure 1-28 Deficiencies in management of public investment ................................................... 62 Figure 2-1 Public budget execution 2014 (in million soles) ......................................................... 66 Figure 2-2 Spending analysis of the Central Government (2014 – Boston Matrix, Level 6) ....... 67 Figure 2-3 Strategic Sourcing Model ............................................................................................ 68 Figure 2-4 Parameters to define a strategy for savings ................................................................. 69 Figure 2-5 Definition of a strategy for savings ............................................................................. 69 Figure 2-6 Product families of items identified for strategic sourcing ......................................... 72 x Figure 2-7 Potential savings on strategic items implementing procurement strategies ................ 72 Figure 2-8 Potential savings of applying procurement strategies to three products amount to between 57 and 115 million Soles each year ................................................................................ 73 Figure 2-9 Procurement processes for sedan cars by method ....................................................... 74 Figure 2-10 Unit price dispersion in sedan cars (in Soles) ........................................................... 74 Figure 2-11 Estimated vehicle life-cost (in %) ............................................................................. 75 Figure 2-12 Unit price dispersion for SUVs (in Soles)................................................................. 75 Figure 2-13 Unit price dispersion for same type of SUVs – 4x4 (in Soles) ................................. 76 Figure 2-14 Fuel unit price by procurement method (in Soles) .................................................... 77 Figure 2-15 Quantity of fuel procured in 2015 and average unit prices by department ............... 77 Figure 2-16 Unit costs for the procurement of cement in 2015 (soles) ........................................ 79 Figure 2-17 Average unit price and percentage bought ................................................................ 80 Figure 2-18 Distribution of the procurement of cement by region (in Soles) .............................. 80 Figure 2-19 Unit price by procurement method and number of bids (in Soles) ........................... 81 Figure 2-20 Procurement implementation timing during 2015 .................................................... 82 Figure 2-21 Procurement timeframe during the year by level of government ............................. 82 Figure 2-22 Length of procurement process by method (in days) ................................................ 83 Figure 2-23 Number of entities that bought these families and percentage of the total procurement (2015) ............................................................................................................................................ 84 Figure 2-24 Number of suppliers and diversification index (2015) ............................................. 84 Figure 2-25 Example on levotiroxina ........................................................................................... 85 Figure 2-26 Number and amount of procurement processes based on method (in soles) ............ 85 Figure 2-27 Number of procurement processes based on the amount .......................................... 86 Figure 2-28 Average number of bids by procurement method ..................................................... 86 Figure 2-29 Processes failed by level of government ................................................................... 87 Figure 3-1 WSS Capital Spending ................................................................................................ 93 Figure 3-2 WSS Spending by Level of Government .................................................................... 94 Figure 3-3 WSS Spending by Source of Financing ...................................................................... 94 Figure 3-4 WSS Sector Investment, 2013..................................................................................... 95 Figure 3-5 Access to water and sanitation is laggign behind peers countries .............................. 96 Figure 3-6 The Marginal Cost of Expanding Access to Improved WSS Services ....................... 97 Figure 3-7 Share of Approved WSS Capital Budget and Share in Unserved Population, by Region, 2015............................................................................................................................................... 98 Figure 3-8 Share of Modified WSS Capital Budget and Share in Unserved Population, by Region, 2015............................................................................................................................................... 98 Figure 3-9 WSS Spending in % of PIM at different government levels ...................................... 99 Figure 3-10 Operating Loss (S/ millions) ................................................................................... 102 Figure 3-11 OCCR in 2014 and 2014 ......................................................................................... 102 Figure 3-12 Hidden subsidies (S/ millions) ................................................................................ 106 Figure 3-13 Investments (S/ millions) ........................................................................................ 106 Figure 3-14 Average Residential Water Rates in Latin America and the Caribbean per m 3 ..... 106 Figure 4-1 Total health spending, % of GDP ............................................................................. 114 Figure 4-2 Health spending between 2009 and 2016.................................................................. 114 Figure 4-3 International comparison of the total spending in health, % of GDP, 2014 ............. 115 Figure 4-4 International comparison of out-of-pocket expenses in health, 2014 ....................... 115 Figure 4-5 Infant and under-5 mortality rate .............................................................................. 116 xi Figure 4-6 Population (1997 vs. 2015) ....................................................................................... 118 Figure 4-7 Health insurance (2006-2015) ................................................................................... 119 Figure 4-8 Service gap: MINSA/GORE subsystem and other subsystems ................................ 119 Figure 4-9 Number of outpatient and inpatient (hospitalization) centers, by network of providers ..................................................................................................................................................... 121 Figure 4-10 Per person spending of the EsSalud and MINSA/GORE/SIS systems ................... 122 Figure 4-11 Discrepancy: SIS coverage rate, by source of information, 2014 ........................... 123 Figure 4-12 Transfer of the SIS capitation payment by insured, 2014 ....................................... 125 Figure 4-13 Bidders in processes to purchase medicines, equipment and supplies, 2015.......... 127 Figure 4-14 Price and quantity of antibiotics and anticonvulsants ............................................. 127 Figure 5-1 Public budget for education and rate of execution, 2011-2015 ................................ 138 Figure 5-2 Public spending as a percentage of GDP, by type of service, 2009-2015 ................ 138 Figure 5-3 Public expenditure by student, by level of education, 2012-2014 ............................ 139 Figure 5-4 Comparison based on distribution of the expenditure by level of education, 2012-2014 ..................................................................................................................................................... 140 Figure 5-5 Expenditure on wages as a % of the total spending in public schools (IIEEs), 2012- 2014............................................................................................................................................. 141 Figure 5-6 Expenditure on personnel, obligations as a % of GDP, by educational level, 2009-2015 ..................................................................................................................................................... 142 Figure 5-7 Current expenditure on educational material in Basic Education as a percentage of GDP, by type of educational material and modality, 2011-2015 ......................................................... 143 Figure 5-8 Capital expenditure as a % of the total expenditure in public schools, by educational level ............................................................................................................................................. 144 Figure 5-9 Regional average expenditure by student in UEs of DREs and UGELs, 2015......... 145 Figure 5-10 Participation of government levels in the public expenditure on education, by region ..................................................................................................................................................... 146 Figure 5-11 Expenditure by student in primary school and learning achieved in mathematics, by UE, 2015 ..................................................................................................................................... 147 Figure 5-12 Percentage of primary school students with a satisfactory level in ECE, 2007-15. 148 Figure 5-13 Performance of secondary school students in reading and mathematics, PISA, 2015 ..................................................................................................................................................... 148 Figure 5-14 Public budget in Education and rate of execution by level of government ............ 149 Figure 5-15 Efficiency frontiers of the public expenditure in primary education in mathematics, TERCE ........................................................................................................................................ 150 Figure 5-16 Efficiency frontiers of public spending in secondary education in mathematics, PISA ..................................................................................................................................................... 150 Figure 5-17 Efficiency frontiers of the regional expenditure on primary education, comparing 2011 and 2015 ...................................................................................................................................... 152 Figure 5-18 Learning achievements if the regions performed at the net efficiency frontier ...... 154 Figure 5-19 Efficiency and lag in learning by UE, based on geographical dispersion, ECE 2015 ..................................................................................................................................................... 155 Figure 5-20 Evolution of the expenditure on highly-effective interventions, 2012-15 (% of GDP) ..................................................................................................................................................... 159 Figure 6-1 International comparison of public spending in citizen security .............................. 165 Figure 6-2 Evolution of real expenditures by administrative government levels ....................... 166 Figure 6-3 Public expenditure on citizen security, by economic classification ......................... 167 xii Figure 6-4 Average rate of homicides per 100,000 inhabitants at the regional level (2000-2013) ..................................................................................................................................................... 169 Figure 6-5 Costs of violence in the private sector – Peru ........................................................... 170 Figure 6-6 Impact of spending on homicides ............................................................................. 175 Figure 6-7 Impact of spending on victimization ......................................................................... 175 Figure 6-8 PNP efficiency on the global level ............................................................................ 177 Figure 6-9 Impact of efficiency on victimization ....................................................................... 183 Figure 6-10 Impact of efficiency on perceptions of the PNP ..................................................... 183 Figure 6-11 Impact of police efficiency on perceptions regarding PNP performance ............... 183 Figure 7-1 Transport budget as a share of GDP ......................................................................... 189 Figure 7-2 Evolution of Transport Sector Spending, in percent of GDP.................................... 190 Figure 7-3 Expansion of the roads network in thousands of Km. .............................................. 190 Figure 7-4 Transport expenditure by type of network, share of total expenditures (%) ............. 190 Figure 7-5 Transport expenditure by level of government, Percent of GDP .............................. 190 Figure 7-6 Current versus Capital Expenditures in secondary and tertiary roads ...................... 191 Figure 7-7 Tertiary Network Condition: ..................................................................................... 191 Figure B.7-8 Institutional framework for the Transport Sector .................................................. 192 Figure 7-9 Evolution of the primary network ............................................................................. 193 Figure 7-10 Peru Road Network ................................................................................................. 193 Figure 7-11 Peru and the Quality of Infrastructure Index: Overall and by sub-Sector............... 194 Figure 7-12 Original and modified budget allocations for Road Sector, National Government 196 Figure 7-13 Original and modified budget allocations for capital spending in the Road Sector, National Government .................................................................................................................. 196 Figure 7-14 Budgeted and Accrued Capital Spending in the Road Sector, Regional Government ..................................................................................................................................................... 197 Figure 7-15 Budgeted and Accrued Capital Spending in the Road Sector, Local Government . 197 Figure 7-16 Budgeted and Accrued Current Spending in the Road Sector, Regional Government ..................................................................................................................................................... 198 Figure 7-17 Budgeted and Accrued Current Spending in the Road Sector, Local Government 198 Figure 7-18 Original and modified budget allocations by type of road ...................................... 198 Figure 7-19 Budget execution of modified budget by type of road ............................................ 198 Figure 7-20 Accrued and Planned Capital Expenditures for Urban Transport, Regional Governments ............................................................................................................................... 200 Figure 7-21 Accrued and Planned Capital Expenditures for Urban Transport, Local Governments ..................................................................................................................................................... 200 List of Tables Table 0-1. Summary of policy options and their impact on efficiency, equity, and potential fiscal savings*......................................................................................................................................... 18 Table 1-1 Revenues, expenditures and fiscal balance .................................................................. 39 Table 1-2 Classification of expenditures by degree of rigidity..................................................... 50 Table 1-3 Analysis of variance and covariance of public spending, by generic expenditure account ....................................................................................................................................................... 51 Table 1-4 Assumptions of the baseline scenario........................................................................... 52 Table 1-5 Overview of policy options and their impact on efficiency, equity, and fiscal savings 57 xiii Table 2-1 More efficient procurement strategies yield fiscal savings of up to 0.4% of GDP ...... 71 Table 3-1 Progress in Management Indicators as per National Sanitation Plan: EPS Only ......... 96 Table 3-2 Approved (PIA) and Modified Budget by Type of Government ................................. 99 Table 3-3 Project Fragmentation ................................................................................................ 101 Table 3-4 Marginal Investment Costs in WSS Coverage ........................................................... 104 Table 3-5 Policy options and potential savings .......................................................................... 108 Table 3-6 Policy options and potential savings .......................................................................... 111 Table A.3-7 Organismos del Gobiernos Nacionales, Regionales and Locales que intervienen en el Sector de Agua y Saneamiento ...................................................... Error! Bookmark not defined. Table 4-1 Budget executed by product unit ................................................................................ 125 Table 4-2 Examples of the allocated and executed budget of budgetary programs ................... 126 Table 4-3 Policy options to improve efficiency and equity and generate savings ...................... 133 Table 5-1 Possible impacts of interventions with incentives and rewards for a good performance ..................................................................................................................................................... 156 Table 5-2 Possible impacts of pedagogic interventions .............................................................. 157 Table 5-3 Impact and potentially scalable interventions by taxonomy of the effectiveness ...... 158 Table 5-4 Blinder-Oaxaca decompositions of the change in learning between 2011 and 2014 . 160 Table 5-5 Options to enhance the efficiency of expenditure in the education sector ................. 162 Table 6-1 Budget Programs in Citizen Security ......................................................................... 172 Table 6-2 Regression Analysis of Budget Allocation Variables (2013)..................................... 173 Table 6-3 Execution of expenditure on citizen security ............................................................. 174 Table 6-4 Inputs and outputs for the efficiency analysis ............................................................ 176 Table 6-5 Regression analysis of the determinants of PNP (in)efficiency ................................. 181 Table 6-6 Optimal values of inputs and monetary quantification in savings ............................. 182 Table 6-7 Impact of savings on the budget and GDP ................................................................. 182 Table 6-8 Reform Options to Enhance Efficiency of the Spending in Security ......................... 186 Table 7-1 Share of Urban Transport Expenditures: the top 5 cities, 2005-15 (in %) ................. 199 Table 7-2 Overall per Kilometer Maintenance and Construction Cost Estimate ....................... 203 Table 7-3 Comparison of annual road maintenance and rehabilitation costs ............................. 204 Table 7-4 Economic Comparison of Budget Scenarios .............................................................. 204 List of Boxes Box 1-1 Peru: Fiscal stabilizers .................................................................................................... 41 Box 1-2 Peru: Fiscal rules ............................................................................................................. 51 Box 3-1 Water Sector Institutions in Peru .................................................................................... 92 Box 7-1 Transport sector institutions.......................................................................................... 192 Box 7-2 Recomendaciones en Planificación del Transporte Urbano – Movilidad y Transporte Urbano Sostenible .......................................................................... Error! Bookmark not defined. List of Maps Map 1-1 Contribution to the total increase in regional government spending, 2015/2012, percentage points .......................................................................................................................... 47 Map 2-1 Average unit price by department (in Soles) .................................................................. 80 Map 5-1 Net efficiency of public spending in education with respect to learning ..................... 153 xiv Map 6-1 Per capita spending in security at the departmental level ............................................ 167 Map 6-2 Evolution of the homicide rate per 100,000 inhabitants at the departmental level ...... 168 Map 6-3 Output Efficiency of the PNP....................................................................................... 178 Map 6-4 Input Efficiency of the PNP ......................................................................................... 178 Map 6-5 Output Efficiency of the PNP....................................................................................... 179 Map 6-6 Input Efficiency of the PNP ......................................................................................... 179 Map 7-1 Peru is falling behind in accessibility, affordability, and duration of transport ........... 195 xv Introduction 1. This Public Expenditure Review (PER) was prepared at the request of Peru’s Ministry of Economy and Finance (MEF). The PER offers analysis, options for reform, and measures in selected areas and sectors to achieve efficiency gains and fiscal savings (where possible), while improving (at least not harming) equity. These efficiency gains and savings can be used to support the on-going gradual consolidation process, to generate space to face future spending pressures coming from the population needs for public services, and to be reallocated within the same sectors or across sectors in programs that show significant positive impact. The PER is framed and informed by the fiscal path established in the multiannual macroeconomic framework (MMF) and in line with maintaining adequate debt dynamics. 2. In the context of less favorable external environment, public finance is one of the critical issues that has come to the fore. Low commodity prices, a slow-down in growth of major emerging market economies, and a sluggish recovery of advanced countries have slowed down economic activity and weighed down on the growth and fiscal outlook of emerging economies, particularly in Latin America. Current projections indicate that commodity prices will remain tamed over the near term, requiring countries to boost productivity growth while generating efficiency gains in their fiscal structures. In this context, Peru needs to improve public services to boost its growth potential, reduce spatial disparities, and meet the growing demand for public services from its population, which puts pressures on the size and composition of public spending. At the same time, the country needs to maintain a prudent fiscal stance to be able to face a less benign external environment that in the commodity boom years. 3. The gradual fiscal consolidation process initiated offers a unique opportunity to carry out important fiscal and sectoral reforms. Fiscal responsibility and sound macro-economic management provided a solid foundation for stability and growth in Peru. Within the solid overall fiscal policy framework, some sectors saw double-digit growth in real terms in their budget allocations in recent years. In some instances, these increases reflected progress in critical sector reforms and a stronger focus on results. But in others, they reinforced pre-existing inefficiencies. Going forward, it will be important to: 4. Preserve the hard-won gains of sound fiscal policy. That includes preserving fiscal buffers and low public debt levels in the context of a challenging external environment, growing contingent liabilities, exposure to natural disasters, and growing population demands for services. The fiscal consolidation would need to be supported by an appropriate mix of spending and revenue measures. 5. Make public spending more efficient. Peru has a number of spending areas in which efficiency gains and fiscal savings can be achieved that will enhance service delivery and equity (or at least not hamper it). It also has some sectors in which it spends less that country comparators, but even within these sectors it could achieve efficiency gains with the available resources. This would create additional space to face future spending pressures or/and can generate resources that can be reallocated to underfunded programs within the same sector that show positive impact. 16 6. This report consists of two parts; Part I focuses on the cross-cutting issues and challenges. The macro-fiscal chapter examines revenue and expenditure trends, provides a fiscal sustainability analysis and an assessment of contingent liabilities. It discusses relevant cross- cutting fiscal issues, expenditure rigidities, revenue challenges, and areas to improve the multiannual macroeconomic framework (MMF). The public procurement chapter evaluates procurement practices applied by all government bodies in the budget. It uses the strategic sourcing methodology and data from Peru’s procurement platform which for the purposes of this analysis has been merged with the budget accounting system (SEAF). It identifies significant fiscal savings and efficiency gains, many of which are relatively easy to achieve. 7. Part II identifies areas for efficiency gains and potential fiscal savings in five selected sectors. These sectors include water and sanitation, health, education, citizen security, and transport. They account for a significant share of the budget and represent a substantial share of public service delivery. These chapters provide several approaches and perspectives in looking at technical and allocative efficiency. For example, part of the analysis of public spending in education and citizen security spending is based on productivity curves that allows to infer the scope for efficiency gains. The health chapter constructed detailed statistics on unit costs to analyze the efficiency of spending in key services. The analysis for water and transport relied on a review of spending trends and outcomes, including an analysis of approved, modified, and executed budgets. 8. The PER identifies at least 1.4 percent of GDP in fiscal savings that arise from efficiency reforms and measures across the areas examined (Table 1.1). The available data allowed to quantify potential savings from the implementation of selected policy reforms, but these are by no means exhaustive. The potential fiscal savings should thus be regarded as a lower bound. Some of savings can help in the consolidation process. But some others may need to be re-allocated within the same sectors (or across sectors) as argued in some chapters. Some of the proposed measures can be achieved without changes in legislation, for example, some of the gains in public procurement, while others require more complex and comprehensive reforms. All of them are feasible over the medium-term, but some may require additional technical work. 17 Table 0-1. Summary of policy options and their impact on efficiency, equity, and potential fiscal savings* Potential Savings (% Sector Measure Efficiency Equity of GDP) Contingencies: explicitly assign all risks and mitigation instruments associatd with new + Neutral PPP contracts to public or private actors Fiscal Budget: reduce variability of the PIA versus the PIM and allocate a larger share of the + Neutral budget at the beginning of the year Options to imrpove the efficiency of public procurement: + Neutral - Full reform of the procurement system (aggressive approach) * + Neutral Public Procurement - Full reform of the procurement system (conservative approach) * + Neutral 0.4% - Reform of 9 low-value and low-risk categories (aggressive approach) * + Neutral - Reform of 9 low-value and low-risk categories (conservative approach) * + Neutral Optimization of human and physical resources at the district level + + 0.3% Citizen Security Creation of the resource allocation formula with efficiency measures + + TBD Funding of programs for the construction of police infrastructure + Neutral Reduce the undervaluation of water and sanitation services for EPS + Improve the EPS' collection efficiency + 0.2% Reduce the unaccounted water losses in the areas in which the EPS operate + Improve planning and coordination of the total investment at the national and + + 0.2% suvbnational levels Water and Sanitation Evaluate the viability of each project as part of a greater investment program to reduce the + + transaction costs for project preparation and implementation Address incentives structures in the budget process that 7undermine efficiency: (i) TBD decouple the budget process from investments to enable multiannual planning of + investment projects, and (ii) change the rules that create adverse incentives Multiannual planning of transport works + + Transport Implement the multiannual budget with MINTRA as the coordinating unit for different TBD + Neutral levels of government Continue to expand the population coverage of insurance Neutral + -- Redefine the package of benefits, keeping in mind fiscal and equity implications + + Adapt international payment systems for suppliers to incentivize the efficient services + Neutral TBD delivery Health Reconfigure specialized hospital care programs + Neutral Improve efficiency in the procurement and distribution of medicines + Neutral see public procurement Build the institutional and operational foundations for the interoperability of insurance + Neutral 0.1%-0.2% Unify complementary services at the regional level + Neutral Remove the implicit subsidies for EsSalud + Neutral 0.1% Provide adequate financing of current expenditures for educational services and integrate + + them into the results-based budgeting framework Introduce the multiannual allocation of investment budgets according to the priorities and strategies of the National Plan for Educational Infrastructure (PNIE); budget the necessary + Neutral operating expenses Prioritize the expenses in highly effective and pro-poor interventions and policies + + Education Implement an electronic government that connects schools, optimizes information flows, TBD + + reduces transaction costs, and speeds up administrative procedures Establish differentiated strategies (by region, university) and accredit capacities at the + + school levels and education units in order to enhance budget responsibilities Establish the conditions to allocate subnational budgets through student-based formulas + + that include equity criteria; start withaccredited schools and education units Link the budget to performance-driven incentives + Neutral Source: World Bank staff estimates. *Note: Potential savings are lower bound estimates. PART I Cross-cutting Issues Fiscal Framework 9. Unlike many countries in the region, Peru saved a significant share of its windfall from the commodity boom years and strengthened its macroeconomic buffers. Fiscal surpluses prior to 2013 kept public debt at around 20–25 percent of GDP, well below the public 18 debt ceiling of 30 percent of GDP established in the country’s fiscal rules. 1 The Fiscal Stabilization Fund’s resources increased from US$100 million in December 2000 to US$8.3 billion (4.5 percent of GDP) in December 2016. Sizeable fiscal buffers provided space to pursue an accommodative fiscal policy during the growth slow-down and are necessary to protect Peru against the fiscal impact of natural disasters, commodity price shocks and the realization of contingent liabilities in the future. 10. The slowdown in economic activity and revenues due the decline in commodity prices, combined with expansionary fiscal measures, led to widening fiscal deficits starting in 2014. Copper and gold account for 35 percent of Peru’s total exports and about 10 percent of total government revenues. As commodity prices declined, economic growth decelerated from an average of more than 5 percent between 2001 and 2013 to 2.4 and 3.3 percent in 2014 and 2015, respectively. The decline in revenues was particularly sharp since 2014 when total (tax) revenues declined from 22.3 percent of GDP to an estimated 18.6 percent in 2016. Almost half of this decline was due to lower revenues from corporate income taxes (CIT). The decline in CIT revenues is explained by the decline in commodity prices reducing the profits of mining companies—which account for one-third of all CIT revenues—and the reduction in the CIT rate from 30 to 28 percent in 2014/15. 11. The dependence on commodity exports resulted in a high volatility of revenues . Between 2001 and 2014, the direct contribution of mining revenues accounted on average only for 13 percent of total tax revenues. But they had a significant impact on revenue volatility—mining revenues accounted for more than half of the total current revenue increase from 2004-12 and for almost 60 percent of revenue volatility in that period. Between 2005 and 2012, current revenues grew at an average rate of 10.7 percent—four percentage points faster than GDP growth in that period. Between 2012 and 2016, current revenues grew on average only 1.5 percent per year, two percentage point slower than average GDP growth. 12. Perú would need to implement a public spending rationalization of 1 percent of GDP over the next 5 years under the baseline scenario to reach the MMF fiscal deficit target of 1 percent in 2021. However, different macroeconomic stress scenarios can raise that number to between 1 and 3 percent of GDP. The baseline scenario assumes strong growth of around 4.2 percent over the medium-term and a gradual increase in commodity prices. Total revenues are projected to increase by 0.4 percent of GDP, requiring a reduction of non-financial public expenditures by 1 percent. Under these conditions public debt peaks at 26 percent of GDP in 2019. In the absence of an expenditure adjustment, public debt is projected to exceed the legal limit of 30 percent of GDP by 2021. But a combination of lower growth and export prices, higher interest rates, a currency devaluation, and lower VAT revenues would push public debt above the 30 percent ceiling by 2019, unless compensated by additional fiscal consolidation measures. 13. Contingent liabilities, which have increased with PPPs, imply additional risks and may increase the size of the required fiscal adjustment. Contingent liabilities associated with PPPs have been estimated to amount to 2.8 percent of GDP in 2015. They largely stem from 1 Peru’s Fiscal Responsibility and Transparency Law also includes the following elements: (1) the non-financial level of spending must be consistent with the structural balance target; and (2) the growth rate of wage and pensions spending cannot exceed the nominal growth rate of potential GDP, which have also been implemented. 19 explicit government guarantees in PPP contracts, for example, for credit risk (0.7 percent of GDP) or minimum revenue and debt service guarantees (2.1 percent of GDP). Moreover, addenda to existing contracts almost always involve additional contingent liabilities. It has been estimated that the contingent liabilities for two large selected projects in Peru might easily be twice as high as initially estimated, highlighting the need for a more detailed quantification of associated fiscal risks.2 A full implementation of the current legislation would help to contain these risks. For example, the government implicitly carries the construction and financial risks associated with PPPs which—according to international best practice—should be carried entirely by the private sector. Moreover, current regulations do not impose an upper limit for annual PPP commitments and government co-payments are typically pre-paid instead of being paid at the time of the service delivery. 14. Despite Peru’s relatively small government size, there is significant room for efficiency gains. With a public spending to GDP ratio of 21.2 percent in 2016, the size of Peru’s government is relatively small compared internationally. Yet, capital expenditures accounted for 5.5 percent of GDP in 2015, which is well above the average for regional (3.0 percent of GDP) and structural (3.1 of GDP) peer countries. Recurrent spending, though low compared to peer countries, has increased rapidly, largely due to i) the implementation of significant education and safety net reforms; ii) an expansion of health coverage; and iii) higher spending on citizens’ security.3 The majority of recurrent expenditures in Peru are rigid (93 percent of total) while the majority of capital expenditures are relatively flexible (only 11 percent are rigid). As a result the share of rigid expenditures (e.g., social protection programs, interest payments, compensation of employees, and pre-committed payments for PPPs) has increased from 66 percent of total expenditures in 2012 to 70 percent in 2015. In this context, the challenge is to balance fiscal consolidation against progress towards closing Peru’s infrastructure gap and increasing resources to improve the quality of public services in social sectors. 15. Given Peru’s needs, the fiscal consolidation cannot solely rely on creating fiscal space through higher spending efficiency. The PER identifies fiscal savings from efficiency gains in public expenditures of at least 1.4 percent of GDP (for the sectors examined). But they will not be sufficient to achieve the desired fiscal path in the case that adverse economic shocks weaken the economy. Thus, Chapter 1 (Macro-Fiscal) also discusses broadly revenue-enhancing measures, even though an in-depth analysis of tax reform options is beyond the scope of this report.4 There is room to raise tax revenues without increasing rates to support the fiscal consolidation. VAT revenue productivity is below the average in the region and OECD countries, which is to some extent due to generous tax exemptions. Eliminating those tax exemptions would increase revenues typically without evidence for an adverse social impact. Tax incentives under multiple tax regimes for different firm sizes are costly but have so far not promoted formalization. Excise taxes on goods with negative externalities are below 1 percent of GDP in Peru relative to 1.8 percent of GDP in LAC and 2.7 in OECD countries. Property tax revenues are at 0.2 percent of GDP in 2013 among 2 Arup (2016), Mejora del Mercado de APP en el Perú, Revisión Ex Post de Proyectos APP, marzo de 2016, mimeo. 3 The increase in current spending increased the share of rigid expenditures (e.g., social protection programs, interest payments, compensation of employees, and pre-committed payments for PPPs) from 66 percent in 2012 to 70 percent of total expenditures in 2015. The majority of current expenditures are considered rigid (93 percent of total) while the majority of capital expenditures are relatively flexible (only 11 percent are rigid). 4 A detailed analysis of the policy options to increase tax revenues without compromising equality, formalization, and productivity would, however, require a more in-depth analysis which is beyond the scope of this report. 20 the lowest in the region, in part, due to the absence of adequate municipal and fiscal cadasters. Better enforcement of VAT and CIT collection could also significantly increase revenues. Finally, encouraging formalization by compensating reductions in payroll contributions, compensated through higher indirect taxes, could also lead to higher and more sustained revenues over time. 5 16. Strengthening budget processes and multiannual planning would improve fiscal efficiency. Large disparities between approved, modified, and executed budgets create large inefficiencies. In 2015, the modified budget was 17 percent larger than the original budget for the central government and 34 percent larger for local governments. These large disparities result from a piecemeal budget allocation process which encourages line ministries to compete for scarce resources by requesting allocations beyond their spending capacity at the beginning of the year. In the second half of the year, line ministries tend to transfer resources not spent to subnational entities to avoid under-execution, undermining planning and leaving local governments with only a few months to spend significant resources. As a result local government budget execution rates are low and about two-thirds of the executed budget is spent in the second half of the year. The budget process also leads to significant spending inefficiencies by delaying the preparation and execution of projects, channeling resources to non-priority areas, and fostering project fragmentation. The introduction of detailed multiyear budgeting and multiannual ceilings for sector spending, aligned with national and sector priorities and the macroeconomic projections in the MMF, would allow to channel more resources to priority areas with more efficiency. 17. Shortcomings in public investment management and the fiscal decentralization system undermine the efficiency of public infrastructure investments. Peru uses standard methods for project selection and appraisal, but does not publish projections of capital spending beyond the current budget year. As a result, there are no multiyear targets or ceilings on capital expenditure by ministry or program. Peru also stands out among its peers in that about half of its capital expenditures are carried out by subnational governments. But high local fragmentation and the current transfer system provides little or no incentives to create service delivery platforms or infrastructure projects with economies of scale that serve several jurisdictions at the same time to improve efficiency. Many municipalities are too small to have the scale for service delivery and, their institutional and technical capacities are limited. The central government has put in place additional procedures and control mechanisms, including the creation of 110 investment committees, which increased administrative and coordination costs. Moreover, the existing criteria for distributing resources among municipalities has led to a regressive inter-governmental transfer system. A reform option could be converting ordinary resource transfers ( Recursos Ordinarios) from the central government to the directorates of regional line ministries into regular regional transfers (the FONCOR), which are allocated in line with fiscal capacity and spending needs, an thus benefit poorer regions with higher spending needs and make subnational government spending more progressive and efficient. Similar improvements could be achieved by integrating the transfer of mining receipts (Canons) into regular local government transfers (the FONCOMUN). Public procurement 5 Fernandez and Villar 2016, “The Impact of Lowering the Payroll Tax on Informality in Colombia”; Rocha, Ulyssea and Rachter 2016, “Do Lower Taxes Reduce Informality: Evidence from Brazil”. 21 18. The modernization of Peru’s public procurement strategy could generate significant fiscal savings and improve public services through lower prices and better quality . The general government procures goods, works, and services worth thirty billion soles in 2014, which is equivalent to 5 percent of GDP. More efficient procurement strategies, could yield up to 0.4 percentage points of GDP in fiscal savings over the medium term. Achieving these savings requires reforms in the procurement system for all product categories purchased by the general government. The report presents two reform scenarios. Under the low-hanging-fruit scenario, procurement strategies would leverage the purchasing power of the government; introduce quality standards to attain lower prices; and increase the use of competitive methods. Under the more comprehensive reform scenario, changes to the procurement strategy also include the adoption of optimization criteria to reduce current consumption levels and the substitution of expensive inputs with less costly but qualitatively equivalent inputs. Fiscal savings opportunities range from 1.1 and 2.2 billion Soles under both scenarios. Moreover, the analysis shows that small adjustments in procurement strategies, which can be implemented without any modifications to current procurement laws or systems, for only nine standardized goods would yield significant fiscal savings each year. A detailed strategic sourcing analysis is provided for three of these goods: vehicles, cement, and fuel. It reveals a large fragmentation of procurement processes and methods, which lead to a large dispersion of prices for the same good paid by different government bodies (sometimes even purchased from the same supplier). For example, prices for cement purchased by different government bodies in 2015 varied by 350 percent; prices for sedan (cars) varied by 600 percent as agencies purchased very different vehicles for the same purpose and almost three- quarters of the procurement processes had only 1 bidder. 19. The public procurement system in Peru is highly decentralized and almost all processes fail in a first attempt, allowing for less competitive methods thereafter. There are more than 2,800 purchasing entities, half of which purchased small amounts of less than 1.9 million soles for an entire year. Ninety-six percent of procurement processes performed during 2015 failed in a first attempt. This leads to delays and allows governments to use less competitive procurement methods in the second attempt. In fact, 42 percent of the total amount purchased in 2015 was contracted during the last quarter and about a quarter of the amount was contracted in the following year. Some of the procured items were provided by a limited number of suppliers and in some cases by only one, suggesting lack of competition. In fact, almost 80 percent of the procurement processes in 2015 had less than three bids. As in many other countries (Chile being an exemption), access to government contracts and procurement in Peru requires that the bidding company has a local presence in the country (e.g., a local subsidiary), providing a de facto entry barriers for (smaller) foreign providers. 20. This section identifies policy options: i. Short term policy options (low-hanging-fruit scenario). It is critical to improve procurement planning to break the administrative silos of individual procurement agencies at all levels of government. This includes carrying out an analysis of the annual demand for each product category to consolidate most of the procurement to larger contracts to exploit economies of scale and reduce the unit prices paid. This can be facilitated by creating appropriate mechanisms to enable cross-regional collaboration. Moreover, the large current volume of uncompetitive awards must be limited; the application of 22 uncompetitive procedures must be the exemption and not the rule. The application of the recently introduced electronic auction modality, for example, has resulted in better participation receiving a higher number of bids and needs to be expanded. Competition can also be promoted by streamlining requirements to prepare a bid and by delivering bidding invitations to different types of suppliers. For vehicles, bids should be evaluated on the basis of price per kilometer, taking into account as maintenance and fuel consumption. ii. Medium term policy options (more comprehensive scenario). To facilitate the consolidation of procurement, regulations should make it legally mandatory for agencies to follow centralized approaches, unless they can justify and prove that they were able to get a better deal on their own. And binding standardized technical specifications should be introduced for all procured product categories. Moreover, the government should introduce appropriate policy requirements and a justified business need before an agency can even start a procurement process. For instance, 4,000 passenger vehicles and 2,000 motorcycles were added to the government fleet in 2015, increasing total maintenance and fuel costs. The cost of leasing goods or services should be considered and evaluated against procurement and maintenance costs. PART II: Sector Analyses Water and sanitation 21. Peru has among the highest public spending in water and sanitation in the region but access to piped WSS services remains below that of its peers, especially in rural areas, and the poor sometimes have to pay significantly higher prices for drinking water. About half of Peru’s rural population does not have access to improved water and sanitation services. This is significantly below access in structural and regional peers. Lack of access is not the result of limited investment. On the contrary, Peru spent 0.8 percent of GDP on average between 2010 and 2014, with a surge in spending that started in 2005. This exceeds the spending of most of Peru’s structural and regional peers. But investment has not translated everywhere into improvements in access. Overall, access to piped water stood at 88 percent in 2015, but only 72 of the total population had access to improved sanitation. And spatial disparities in access to both water and sanitation are still significant—only 60 percent of rural households have access to water and only 22 percent to improved sanitation. There is also a large gap between urban, peri-urban, and rural areas in the quality of basic service provision. As a result, households in fast-growing peri-urban areas often pay much higher fees: drinking water can cost up to 6 times more in informal settlements than in in residential areas where it is available as piped water. 22. The sector suffers from deep-rooted inefficiencies and hidden subsidies that have weakened the financial performance of water utility companies (EPS). Consumer subsidies amount to about US$350 million or 0.2 percent of GDP per year. They arise from the mispricing of services, collection inefficiencies, and non-revenue water (NRW) losses. The mispricing of services in the current tariff regime provides hidden subsidies to existing consumers. The NRW losses, on the other hand, provide hidden subsidies to unconnected consumers. All these inefficiencies pose a claim on public resources and discourage the provision of appropriate operations and maintenance (O&M) spending for existing infrastructure. Most EPS, including 23 SEDAPAL and other large utility providers, were unable to increase tariffs in line with their increases in their O&M cost, leading to a decline in their average operating margin from 28 percent in 2005 to 10 in 2014 and generating operating losses for some. 23. The financial position of the EPS leads to a vicious circle: low tariffs, lack of maintenance spending, weak EPS management capacity, insufficient local service quality, and bail-outs from the central government adding to the fiscal costs. The reliance on (additional) transfers or occasional central government bail-outs to cover O&M costs, discouraged more efficient management of the EPS, further weakening their capacity. The lack of own revenue capacity led to insufficient maintenance spending, deteriorating the infrastructure capital stock and raising future rehabilitation costs. As a result, the average hours of service provision per day slightly declined from 17 days in 2005 to 16.5 in 2014 while metering and active connections stagnated since 2005, falling significantly short of key targets. And it has pushed consumers to build water storage facilities in their houses and has forced those without network access to pay up to 6 times higher prices for clean drinking water from private vendors. The low quality of services, in turn, made it more difficult to justify local tariff increases. 24. The cost of each additional connection tends to increase exponentially as access approaches universal coverage, but this cost increase has occurred much earlier than expected in Peru pointing out to potential inefficiencies. The last consumers without connection tend to reside in more remote or isolated areas, which can significantly increase the cost of making the additional required connections. The geographic and demographic characteristics of each country influence the marginal cost of additional connections. In countries with close to universal access, the unit cost of connecting each of the last unconnected households in remote areas is high due to the difficult geography. In Peru, it is estimated that the marginal cost of each additional connection would increase exponentially when approximately 85 percent of the water supply coverage is achieved – that is, significantly earlier than in other countries, even those with challenging geographic settings. This is typically an indication of inefficiencies in investment. 25. Investment projects are not aligned with sector priorities. The distribution of the originally approved budget for WSS capital spending has a weak correlation with local WSS service needs (population without improved WSS service). This is in part due to way capital spending is funded—the main sources are ordinary resources ( recursos ordinarios) and canon transfers. These resources are not equally distributed across the country and disproportionally benefit the few municipalities and regions that have large mines. Mid-year budget modifications, however, had a stronger positive correlation with WSS service needs. But the large mid-year budget modifications have also led to other problems as discussed below. 26. The unpredictability of budget allocations and modifications leads to spending in projects with short term execution (that may be less strategic), project fragmentation, delays, and low execution rates. Large budget revisions are quite common in Peru (see Chapter 1), but they are exceptionally large in the WSS sector. The mid-year modified WSS budgets exceeded the originally approved WSS budgets on average by 50 percent between 2009 and 2015, doubling the average for the complete central government budgets of 25 percent. And the modified WSS budget of local governments even exceed the originally approved WSS budget by more than 400 percent from 2010-14. 24 27. Moving towards universal coverage in water and sanitation by 2021 is unaffordable without addressing the sector’s large spending inefficiencies. The significant increases in spending, which started in 2005, have not resulted everywhere in the anticipated progress. Without realizing efficiency gains, it is estimated that achieving nearly universal access in water and sanitation will cost up to 33.8 billion Soles or almost 1 percent of GDP each year from 2017-2021. This is well above the spending levels over the last decade and would not be affordable. However, these costs would be significantly lower if the government were to implement reforms to improve spending efficiency in the sector to a level observed in other countries in the region. It is estimated that the accumulated cost of moving towards universal coverage would decline by half if Peru would achieve the same level of public spending efficiency in the water and sanitation sector as Brazil or Mexico, which share some similar topographic challenges and higher coverage rates. 28. Improving public investment management and performance of public utility companies alone could lead to fiscal savings of around 0.4 percent of GDP and accelerate the access to and quality of water and sanitation services. The most urgent need for reforms to address the large spending inefficiencies is in three areas: i. Raising capital and operational efficiency. The weak financial performance of the EPS poses a significant burden on the central government’s budget and a heavy cost on consumers. Raising the efficiency of EPS to remove the hidden subsidies would save at least 0.2 percent of GDP each year. It requires the use of water pricing as a tool to improve efficiency while safeguarding the affordability of access. As the sector is highly decentralized, this will require strong coordination between the different stakeholders in the sector. It also requires a communication strategy as the weak quality of services in the past has created substantial discontent in the population. Moreover, it is critical to align the incentive system for EPS management with sector spending priorities, including through multi-annual plans that incorporate revenue management, infrastructure asset management, and debt management. At the same time, policy makers should also increase their efforts to improve the planning and technical capacities of municipalities and service providers. ii. Strengthening sector investment planning. The introduction of clear rules and ceilings to allocate funds based on sector priorities at the time of budget approval would allow subnational governments to plan and implement their capital budget with higher efficiency. This includes implementing a multi-annual public investment framework with clear priorities and direct links to the budget. It should include not only the prioritized investments but also the financing of these investments and consider the complementarity between sector investments from municipalities and EPS. For Peru largest EPS (SEDPAL) alone, it is estimated that better investment management—evaluating investment plans more comprehensively and aligning them with sector priorities—would save 25 percent of the investment costs or 0.2 percent of GDP, without compromising services. Improving budget allocations. Better planning as highlighted above will also need to reduce the prevalence of mid-year revisions and thus lead to higher execution rates on more strategic projects that tend to be multi-annual. 25 Health 29. Peru’s transition to an upper middle income country has coincided with significant improvements in health outcomes but has also created new challenges in the health sector. During the past decade, Peru successfully lowered child mortality and reached near-universal coverage of immunization and prenatal care. It introduced a Comprehensive Health Insurance (SIS), which has increased the share of the population with health insurance from 38 percent in 2006 to 73 percent in 2015 and helped expand health sector infrastructure and human resources in poor rural areas. The current challenges of Peru’s health sector are to find the most efficient way to increase health coverage; increase the capacity to deliver health services in poor districts; reduce out-of-pocket spending; and prepare the public health system for Peru’s demographic and epidemiological transition, which is a by-product of urbanization and growing income levels. 30. Improving public spending efficiency in the health sector would not only generate fiscal savings that can be reallocated to higher priority areas within the sector, but also help Peru make progress towards a modern health system. Public health care spending (including health insurance) in Peru is at 3.3 percent in 2016 which is relatively low by international standards. Yet, there exist significant inefficiencies in the sector. Addressing these inefficiencies will be important for Peru to meet its objective of developing a modern and inclusive health system. Spending inefficiencies in this sector are prevalent in four areas. 31. Peru’s fragmented health systems lack interoperability, increase infrastructure and administration costs, lead to cross-subsidies, and leave gaps in access to services. The health system is broadly administered by 3 different entities: the Ministry of Health (MINSA), which provides health services for about 60 percent of the population; the Social Health Insurance EsSalud, which covers around 30 percent of the population; and the Armed Forces and National Police. Despite recent efforts, the interoperability between these systems is quasi non-existent, leading to costly duplications and overlap. As a result, spending on health administration and management accounts for 15 percent of total current health spending, which is significantly higher than in other LAC countries or the OECD average (3 percent). At the same time, the parallel systems create gaps in the access to services. The majority of medical centers not providing hospitalization, for example, are part of the GORE/MINSA network or are private while EsSalud operates few of such ambulatory centers. And the public sector operates a large number of primary care centers in rural areas while EsSalud’s hospital network is concentrated in urban centers. This has also led to implicit cross subsidies provided to EsSalud since public health facilities are open to its beneficiaries but EsSalud is not required to reimburse the public insurer for the services. Efficiency gains from exploiting the complementarities in facilities and services between the different networks would not only reduce administrative and infrastructure costs but also allow to offer more complete and locally affordable health service packages to the population. 32. There mechanisms to prioritize spending are limited. Even though the public health insurance systems, SIS, was designed to apply result-based budgeting, the allocation of health expenditures still appears to be based on historical budgets in practice. A recent law expanded the basic health services package for beneficiaries, but this has not pushed the need to prioritize expenditures. The expansion of basic health services has rather generated rationing of services based on local supply, fostering inequality in access. An excessive number of patients is treated in 26 expensive hospital settings since primary and preventive care cannot keep up with the demand, raising costs unnecessarily. Unlike other countries, Peru does also not use its universal health insurance register to allocate spending more efficiently. So far, the register has only been used to monitor client affiliation but not to facilitate the collection of insurance contributions, to remove overlaps with other social assistance programs, or to expand insurance coverage. Finally, the mechanism used allocate expenditures in EsSalud is also unclear. 33. The health sector financing mechanisms are fragmented and inefficient. There are parallel systems to finance public expenditures for goods and services, for example Financing for Results, Social Insurance, and the public budget, which creates high transaction costs and reduces transparency. The public insurance system SIS introduced per capita spending to finance primary care services. But SIS finances only 12 percent of the total health costs of its beneficiaries, while the rest effectively continues to be financed based on allocations of historical budgets or out of pocket. As a result, the majority of health financing is still based on paying for the inputs claimed by service providers and hospitals (supply) instead of payments based on the coverage or quality of the services provided (demand and outcomes). The introduction of more efficient health financing methods is also limited by the absence of mechanisms to verify the services provided by hospitals, in part, due to the lack of appropriate information systems and auditing. Finally, the procurement of medicaments and medical equipment often results in high prices due to the lack of competition among suppliers as well as the lack of use of more competitive procurement methods (see Chapter 2). 34. Information systems are weak and do not allow to monitor spending. The sector does have fully reliable statistics on basic information such as the total number of health sector employees or the total number of hospital beds. Modern management systems are absent in most hospitals, resulting in inefficient processes and long waiting times for patients. In particular, hospitals and health centers lack IT systems for admission, clinical labs and images, human resources, and electronic prescriptions. The absence of these IT systems makes it impossible to monitor the quality of services—the prevalence of hospital infections, for example, is unknown— or to determine the appropriate level of financing, both creating opportunities for misuse of public funds. The lack of modern IT systems and standards also limits the interoperability between the different health insurance systems. 35. Improving the efficiency of public health care spending requires modernizing the system along the following dimensions: i. Ensuring the full interoperability of the health systems can generate fiscal savings of at least 0.3 percent of GDP. There are several measures that can be taken to allow for more interoperability. This includes improving the medical protocols to treat patients, making the health facilities of the EsSalud network accessible to the beneficiaries of public insurance, and making transparent and addressing cross subsidies between the different systems. Removing the cross subsidies to EsSalud by adequately reimbursing SIS services, for example, would generate fiscal saving of 0.1 percent of GDP. Harmonizing health financing methods and modernizing IT systems would further enable more interoperability. In the medium term, however, the government needs to evaluate whether universal health insurance can be achieved within the current institutional framework of the sector, or 27 whether it is necessary to unify the different fragmented health insurance systems, which other countries at similar stages of development have achieved in the past. In this regard, it is estimated that efficiency gains from harmonizing the fragmented health insurance administrations alone, can generate fiscal savings of up to 0.2 percent of GDP by reducing the administration costs without compromising the quality of health services. ii. Improving the prioritization of spending. In the public system, it is essential to redefine the benefits of the basic package to have clarity on the involved services and the costs of a potential extension of the package. It is also necessary to monitor the beneficiaries to ensure equality in access. In EsSalud, the scope of benefit packages should be redefined according to the revenues collected. Moreover, guidelines should be provided to define adequate health services depending on health conditions, ranging from primary to more complex specialized services. iii. Developing a strategic health financing system. These guidelines should serve as a basis to develop a system of strategic financing, replacing the current passive system. The strategic system should include, among others, (i) participating actively in the prioritization of provided services; (ii) choosing the combination of suppliers that maximizes benefits; (iii) conditioning payment and financing systems on the performance of suppliers; and (iv) introducing co-payments for patients who demand specialized or hospital services for primary care health conditions. The universal health insurance register could provide the foundation to develop the strategic financing system. More generally, it is important to review the rich international experiences in introducing strategic health financing systems. iv. Building a modern health information system. It is critical to invest in the modernization of health information systems in order to be able to track spending and plan and execute health budgets efficiently. Moving towards true interoperability of the health insurance systems requires, for instance, (i) compatible and modern IT systems, (ii) standardized classifications for patients, providers, costing, services, etc.; and (iii) secure health communications network that enables providers, insurers, and beneficiaries to securely transmit and exchange confidential health information. Education 36. Improving the quality of education is critical for Peru to continue on the path of sustained growth. Coverage and quality of Peru’s education has improved rapidly in recent years. Peru achieved the biggest gains on the PISA score of all participating countries between 2009 and 2015. The quality of primary education also increased relative to other countries in the region. Despite the improvement, it remains at the lowest level in South America and the seventh lowest among 73 participating countries worldwide among PISA participants. According to those PISA results half of all students in secondary education cannot read properly or do not comprehend what they read—and almost three-quarters of second graders do not have a satisfactory level in Mathematics. Moreover, the learning outcomes of schools in poorer, more remote areas are significantly worse. In fact, Peru has one of the largest inter-regional performance differences among PISA participants. Recognizing the importance of achieving better learning outcomes, the government has undertaken significant education reforms in recent years. 28 37. Peru spends less than country comparators in education, but it could achieve efficiency gains even at its current level of spending. Between 2011 and 2015, education spending as a share of GDP increased from 3.4 percent to 4.1 percent, driven by an increase in teacher salaries and teaching material. The recent increase in spending on learning materials has contributed significantly to the better learning outcomes. In 2015, education accounted for 17 percent of total spending—the same level as regional peers but slightly higher than structural peers. Student learning outcomes in primary and to a lesser extent also secondary education tend to be better in Peru than in countries with comparable spending per student. But Peru spends less on education per student both, in absolute terms and as a percentage of GDP, than its structural and regional peers. Spending per student, for example, is 2.5 times higher in Mexico and 3 times higher in Chile. While there is a case to spend more over the medium term, there are also significant inefficiencies in the sector that need to be addressed in parallel. The analysis of the sector’s spending efficiency should further guide where (not) to investment potential additional resources. Spending inefficiencies in the educations are prevalent in three areas. 38. High infrastructure spending in education has not contributed to better learning outcomes and resulted in significant spending inefficiencies. The share of capital expenditure in total expenditure in education is substantially higher than in regional or structural peers, which is to some extent explained by a higher unmet demand for school infrastructure. While this is in principle should be good news, the high infrastructure spending has not contributed to better learning outcomes in Peru. And it has resulted in significant capital spending inefficiencies. First, Peru’s spending on maintenance is significantly lower than in regional and structural peers, causing a deterioration of school infrastructure capital stocks and involving higher on-going and future rehabilitation/re-construction costs. Second, the prioritization of investment spending is weak due to the absence of a medium-term prioritization strategy that takes into account equity and efficiency criteria. Third, a significant share of school infrastructure spending in Peru is conducted by local governments which have lower execution rates and planning capacity, further undermining capital spending efficiency. 39. The average expenditure per student by regions and local spending units does not take into account regional and local geographic or socioeconomic challenges. There are remarkable differences in educational spending across regions. They are, however, not necessarily correlated with better learning outcomes. Some regions spent twice as much per student in primary and secondary education in 2015 without achieving better outcomes. Students in Piura, Cusco, and Ayacucho, for example, achieved similar test scores in Mathematics but Piura’s spending per student was 25 percent lower than Cusco’s and 50 percent lower than Ayacucho’s. A more detailed analysis for the local executing spending units within regions shows that only to a limited extent spatial differences in spending efficiency can be explained by spatial differences in crime or poverty rates and geographic distance to local school administrations, among others (see chapter 5). As a result, spending per student is well below what is required to provide quality services in some regions—Loreto and Ucayali are two regions in the Amazon, for example, where the provision of services and the retention of qualified teachers have higher costs, especially considering the need for bilingual education. 29 40. And yet, significant differences in spending efficiency across regions remain even after controlling for regional geographic or socioeconomic challenges . The average expenditure per student by region is only weakly correlated with regional learning outcomes even after taking into account the real local costs of service delivery due to large distances to local school administrations or high local poverty or crime rates. Ancash or Puno, for example, have comparable spending per student, geographic challenges, and even higher poverty rates than Cusco or Arequipa, but their students achieved better learning outcomes. Simulations suggest that if all regions in Peru spent resources efficiently, the share of students with a satisfactory performance in mathematics would increase significantly (for example, by at least 10 percent in primary education). 41. The following policy options can help raising the efficiency of education spending: i. Refining capital investment planning and execution. A first important step to improve the prioritization of investment projects is the adoption of a National Plan for Investment in Education which should set clear criteria of cost-efficiency, maximizing the number of benefitting students balanced with equity criteria. This would allow in a second step for multiannual planning and the collection of information necessary for the preparation of investment projects to minimize transaction costs. It is also critical to integrate maintenance and operating costs into investment planning, already taking these into account when prioritizing investments. As in other sectors, low execution rates in education are also caused by the large budget revisions throughout the year, undermining the planning and investment capacity of local governments and channeling resources to short-horizon, potentially low impact projects. ii. Reallocating spending to more effective programs. Despite recent improvements in budget allocations, for example through the wide-spread implementation of result-based budgeting, allocations do not always target the most effective interventions. Peru has prioritized infrastructure capital spending relative to peer countries in the past without much evidence that the high capital spending contributed to the improvement in learning outcomes. There is convincing international evidence that pedagogical interventions have the greatest positive impacts on student learning outcomes, especially in developing countries. The acquisition and distribution of learning materials has also led to better learning outcomes in developing countries. The recent increase in these expenditures in Peru have also contributed to the significant improvements in learning outcomes, especially among the poor. In the medium term, international experiences also suggest that interventions supporting teachers promise better learning outcomes—these interventions are costly and materialize only gradually over time. iii. Improved mechanisms to assign resources to sub-national governments . Over the medium term, the introduction of per-capita funding formulas, as part of a broader and comprehensive reform of the intergovernmental transfer system in the country, could contribute to improvements in efficiency, equity, predictability of resources and transparency in the sector. But for this reform would require working on a number of pre- conditions that today are not in place. 30 iv. Strengthening information systems. The implementation of systems that allow to monitor the use of available resources by government officials, teachers, parents, and students can reduce inefficiencies. Recent efforts, such as the “Semáforo Escuela” monitoring the performance of student evaluations (ECE) at the school level, could be expanded to achieve better learning outcomes. And portals such as “Ponte en Carrera” are important steps towards improving transparency and helping to design cost-effective interventions. Citizen Security 42. An effective fight against crime and violence in Peru does not always mean spending more, but also spending existing resources more effectively. The level of violence is relatively low compared to the region. Peru had the second lowest average homicide rate between 2000 and 2013 after Chile, among all Latin American countries. But the trend of violent crimes has been increasing rapidly in recent years—especially crimes against property such as robbery and violent robbery—coinciding with the intensification of drug trafficking. As a result, security problems have become one of the top concerns for the population. The increased awareness and demand for citizen security services led to higher government expenditures. Spending on citizen security almost doubled in real terms between 2009 and 2015. The sector accounts for about 9 percent of total expenditures, the fourth largest after education, health, and transportation. The share of total budget allocations for citizen security is considerably higher than in regional and structural peer countries but citizen security spending as a share of GDP is not. 43. The aggregate spending efficiency of the PNP is low compared to structural and regional peer countries. Peru has a relatively low percentage of crimes resolved given its total number of police officers relative to its regional and structural peers such a Romania and Thailand according to several studies. Within the region, Chile and Colombia show much better outcomes for the same resources spent on total inputs such as the police force, vehicles, or information technology. Several studies also estimate that the average spending efficiency of regional PNP departments is lower than in other Latin American countries. 44. There are no effective mechanisms to prioritize spending. The allocation of security resources is generally based on local levels of crime and violence and the size of the population. Nevertheless, departments with a high concentration of crimes have not necessarily received high budget allocations in recent years. Spending is also correlated with some socioeconomic and demographic factors but not others. In particular, higher poverty rates and higher inequality have not led to higher budget allocations in the past, despite their strong correlation with crimes. And departments with larger number of reported crimes have also not received more resources. Similar to other sectors, local governments show low execution rates further undermining the prioritization and efficiency of spending. 45. Reallocating spending across police districts can achieve higher quality services and better security outcomes without increasing the average spending per police district. The PNP could improve the quality of its services, measured by the number of prevented crimes, by 60 percent with the same level of inputs such as the number of police officers, vehicles, and technology (that is, computers and telephony). There is also a large heterogeneity across 31 departments and districts. The departments of Moquegua and San Martin show the highest levels of efficiency measured by the allocation of inputs they received relative to their number of prevented crimes. At the district level, the efficiency of inputs ranges from 4 percent for the district of San Juan de Miraflores to 100 percent for the districts of San Juan de Lurigancho or Vista Alegre. Only 90 districts out of a total of 795 districts (11 percent) with data availability are efficient. The local efficiency of the PNP is affected by socioeconomic factors such as unemployment rates in the districts or the prevalence of alcohol consumption which should be taken into account for the allocation of PNP spending. One important factor in Peru is the local presence of vigilante groups which exist in half of all municipalities in the country. In many cases, these groups patrol jointly with local PNP forces. However, this study finds that the presence of vigilante groups reduces the local efficiency of the PNP, suggesting the duplication of efforts and thus an inefficient allocation of PNP personnel. 46. Reassigning police forces and other inputs across police districts could improve citizen security without raising spending or realize fiscal savings of about 0.3 percent of GDP without compromising the quality of services. The following reform options help materialize the efficiency gains: i. Introducing a formula for budget resource allocations. The allocation of resources for citizen security and in particular the PNP is not based on a formula or methodology. The main priority for such a formula should be a mechanism to distribute police forces to districts and departments where they are needed more urgently. Moreover, funding should not only be based on policing needs, but should also take into account local efficiency criteria. In this regard, the criteria for the reallocation of resources should account for the observed distribution of reported crimes and regional socioeconomic factors such as local poverty rates or levels of inequality. ii. Reassigning inputs and personal. Reducing the number of police officers in districts that show less efficiency in the use of resources (that is where officers are less needed) would bring fiscal savings of 0.27 percent of GDP without compromising the quality of police services and security. The results are based only on district police stations, capturing only about 40 percent of the total PNP staff. The potential gains can be much larger if the efficiency of the police organization as a whole is optimized. Reducing the number of vehicles used in less efficient district police stations would bring fiscal savings of at least 0.04 percent of GDP. The potential gains can be much larger if, in addition, the efficiency of the police organization as a whole is optimized. iii. Improving the coordination between police forces and vigilante groups. While vigilante groups can be effective in preventing certain crimes, the PNP performs many more tasks of control, combat, and crime prevention. Thus, for patrolling and prevention of violence and crime, better coordination of activities between the PNP and vigilante groups would improve PNP’s efficiency and the freed resources could be reassigned to other police tasks. Transport 32 47. Despite a substantial increase in transport expenditures, Peru still faces important gaps in the quantity and quality of its infrastructure. Public expenditures in the transport sector accounted for 2.2 percent of national GDP in 2015, having increased more than four-fold during the last decade. This trend has been accompanied by a significant shift in public expenditure authority and increased allocation of resources to subnational levels. The bulk of the increased expenditures have been devoted to roads and urban transport: almost 90 percent of all transport expenditure are accounted for by these two sectors. Consequently, the size and quality of the primary road network has improved. But secondary and tertiary roads are in much worse conditions: only 12 percent of the tertiary network is in good condition. Moreover, inadequate attention to maintenance raised costs and reduced service quality. Resources for urban mass transport have also increased rapidly. 48. The low quality of Peru’s transport infrastructure limits growth. The density of Peru’s road network averages 0.12km per square km and only 14 percent of the road network is paved, which is significantly less than in structural and regional peers. Moreover, the primary transport network connects Lima to other coastal cities, but not necessarily to medium and smaller cities in the Sierra or Selva, limiting Peru’s economic development and reducing the competitiveness of its agricultural and other sectors. Peru also ranks below regional and structural peers in the quality of transport infrastructure and logistics costs. High transport costs in Peru not only hamper the profitability of exporters but also of producers and service providers for the domestic market. And they increase the costs of access to better quality services for the poor at reasonable prices. High aggregate spending levels and weak sector performance, even after controlling for Peru challenging geography, point to the existence of significant spending inefficiencies. This chapter discusses three areas of transport expenditure inefficiencies. 49. Issues with budget planning, modifications, and execution, generate important inefficiencies. Budget Planning in Peru is guided by its Transport Strategy, but the strategy lacks specificity and a sector-wide transport master plan has not been articulated in detail. The Strategic Sector Plan 2012-2016 for the road sector, prioritizes 22 logistics corridors which carry the bulk of Peru’s commercial and local products but is less clear about the strategic priorities for secondary and tertiary roads. The absence of a master plan also blocks the development of a systematic framework for budget allocations to both, network development and maintenance. Moreover, as in other infrastructure sectors, the large revisions is modified relative to the originally approved budgets, creating spending inefficiencies. The large revisions foster the fragmentation of projects into smaller entities and the prioritization of short-term horizon low impact projects. And for larger, multiannual projects the unpredictability of funding can lead to implementation delays and higher costs both, in terms of capital costs—the cost of construction increases due to these delays— and in terms of recurrent costs—the supervision and implementation support budgets that make up a significant part of the total recurrent costs also increase. Budget unpredictability also weakens the project prioritization process since it reduces the implementation time as modified budgets are approved mid-year. 50. At the sub-national level, the fragmentation and the current transfer system provides little incentives to join forces in inter-jurisdictional transport projects with economies of scale. The trend to finance smaller and more fragmented projects has increased. The number of investment projects at the local level increased from 2,100 in 2004 to more than 15,000 in 2014 33 and the average costs per project shrank dramatically. The large number of small projects suggests that investment decisions are uncoordinated limiting positive spatial spillover effects across local governments. Moreover, municipalities with more abundant resources from the Canon transfers may not have the highest investment needs. 51. Maintenance spending is insufficient, generating higher costs of rehabilitation and re- construction. Maintenance spending at adequate levels tends to have a high economic return and ensures the sustainability of existing assets. Yet maintenance budgets are often too small in the sector in Peru, especially for sub-national governments. Sub-national governments have to comply with the technical standards set by the MTC (e.g., applicable to low-traffic roads) and by MEF (e.g., use of specific intergovernmental transfer for routine road maintenance). But local staff responsible for budget preparation in sub-national governments has been found to be not always familiar with existing technical standards for the periodic and routine maintenance of rural roads. Peru is starting to rely more on road concessions and new initiatives such as the microenterprises for routine maintenance program or the CREMA contracts pilot. The use of these programs should be increased. And more budget resources should be re-allocated for maintenance for subnational roads. 52. The implementation of the new legal framework for PPPs is lagging behind . Peru is still lacking an infrastructure plan and procedures capable of generating a pipeline of mature projects. In addition, the institutional capacity for developing projects is often low. A successful implementation requires also that governments invest heavily in high quality pre-construction studies and surveys, that they structure contracts carefully to achieve an appropriate distribution of risks and expected rewards, and that the mix of private-public investment and involvement be adapted to specific circumstances. 53. Some key areas for reform to address spending inefficiencies are as follows: i. Developing a sector wide transport master plan. A master plan at the aggregate sector level provides the basis to developed more refined sub-sectoral transport strategies. Such a plan needs to reflect priority areas, including issue of multi-modal corridors to reduce export costs. The plan also needs to incorporate maintenance spending needs that emerge from past and current investments to reduce unnecessary rehabilitation and re-construction costs. This can pay off significant fiscal savings over the long run. The current transport strategy could be improved to develop a sector master plan by (i) completing the main national investment corridors; (ii) improving the condition of existing road assets in the secondary and tertiary networks; (iii) ensuring adequate maintenance spending to minimize total road user costs and improved road safety; (iv) involving institutions and agencies at all levels of government in identifying issues and planning road interventions; and (v) ensuring the sustainability of the road programs by prioritizing investments in alignment with available resources while expanding the tolling network to increase the revenues available. For the roads sub-sector, increasing the application of standard road asset management systems, revisiting technical standards, and building capacity at the local and regional level would go a long way in improving the efficiency of the sector spending. In the medium term, simplifying the current complex institutional arrangements would also 34 help streamline the budget approval process and improve predictability of the investment and maintenance funding. ii. Providing a framework for expenditure allocations for maintenance spending. The adoption of adequate mechanisms to allocate resources to network development and maintenance spending is a priority. Permanent maintenance programs addressing both routine and periodic maintenance should be included and supported by performance indicators which help the development of maintenance contracting by level of service. In the short term, it is be important to implement maintenance programs for all roads, even if initially the cost estimates are based on rules of thumb. Once programs are in place, maintenance needs can be assessed more accurately using direct assessment based on the application of standard road asset management systems (and while those are developed using more uniformly across the country indirect assessment based on formulas related to traffic, road length, and other variables affecting maintenance needs). iii. Using PPPs more effectively. Peru has come a long way in attracting private participation, especially in roads and mass transport, and has recently revised its legal framework for development and regulation of PPPs. Introducing a value for money analysis (VfM) to compare if projects under PPP schemes provide savings in comparison to public works, however, remains a priority. The implementation of a multi-annual public investment framework with clear priorities and direct links to the budget would help improve budget planning, modification, and execution which would increase the investment spending efficiency by reducing the uncertainty of project funding. 35 Part I – CROSS-CUTTING ISSUES Introduction 1. One of the most outstanding characteristics of Peru’s economy over the last decade has been its macroeconomic stability. Albeit that Peru’s economic success over the last decade can be attributed in part to the increasing export prices, the price stability, the prudent fiscal stance and the structural reforms made in several economic institutions were also important elements strengthening the competitive advantages of the economy. 6 In a setting with low rates of inflation, positive fiscal results on average, proper management of the public debt, a modest external gap financed with long-term capital, and correct management of the exchange policy, the Peruvian economy grew by almost 6 percent per year between 2006 and 2015. 2. Peru saved part of its windfall revenues during the commodities boom, providing it with macro-fiscal space to respond to external shocks. Throughout the high-growth period largely driven by the high commodity prices between 2004 and 2013, Peru saved a significant part of its earnings, contrary to several other Latin American countries. This was reflected in a domestic savings rate that increased from 15 percent of GDP, at the beginning of the decade, to 22 percent in 2013, which has helped finance a considerable part of the investments in recent years in addition to ensuring the accumulation of solid macroeconomic buffers to face times of hardship. Public savings were a determining factor for this result as well, increasing from 1.8 percent in 2004 to 7.2 percent in 2013. In this context, the public debt fell to around 23 percent of GDP, one of the lowest rates in the world. Furthermore, the Central Bank accumulated international reserves equivalent to around 32 percent of GDP at the end of 2016. 3. The adverse shocks affecting the terms of trade over the last 4 years have had an impact on growth and the fiscal accounts. The terms of trade have displayed a downward trend since 2012, with the deepest point in 2015. This has had a negative impact on a series of variables, such as total investment, the product, and fiscal revenues. This coincided with a series of reforms in education, health and citizen security which had been underway since 2011 and which gave rise to an increase of current expenditure, particularly in wages. Concurrently, and given the magnitude of the external shock, in 2014 the authorities decided to carry out a tax reform, which led to an additional drop of current revenue, beyond the effect of the automatic stabilizers on the revenue side. Thus, the fiscal deficit reached 2.7 percent of GDP in 2016. 4. A less benign and more volatile external context, combined with a growing demand for public services and infrastructure, would entail maintaining the fiscal discipline and improving public spending and the revenue structure. In 2014, the economy entered a structural slowdown process towards a more moderate potential GDP growth compared to the years of the commodity boom. In view of the medium to long-term nature of the external shock, the authorities are planning to enter a period of gradual fiscal consolidation. On the expenditure side, the focus should be on the areas with opportunities to enhance efficiency and generate fiscal 6 World Bank (2015). 36 savings without affecting the provision and quality of public services. Moreover, there are some areas and sectors in which it is possible to reduce inefficiencies and secure medium-term savings in order to plan for the growing pressure for services in the future. On the revenue side, and given the relatively low level of tax collection compared to similar countries, it will be important to analyze tax reform options to improve efficiency of the revenue and collection structure in order to create fiscal space while reducing the negative effects on the allocation of resources in the economy. General Fiscal Dynamics More revenue volatility but with a prudent fiscal management: Historical perspective 2004- 2016 5. Fiscal revenues have been highly volatile over the last decade . Fiscal revenues went from a growth rate of 24 percent in 2006 to a drop of 11 percent during the 2009 financial crisis. Beyond the numbers, it is possible to distinguish two periods based on the average growth rhythm. First, a rapid growth period between 2005 and 2012, when current revenues increased by an average real rate of 10.7 percent, 4 percentage points above the average GDP growth rate. In that context, from accounting to 18.7 percent of GDP at the beginning of the past decade, current revenues reached a peak of 22.8 percent in 2012. The international context with increasing mineral prices was a key factor for this to happen (see Figure 1-1). Second, there was a rapid slowdown between 2013 and 2016, with a negative average growth rate of revenues of 1.2 percent, while GDP growth was 3.9 percent. Figure 1-1 Increase in terms of trade and Figure 1-2 Fiscal revenues from mining, government revenues other sectors, and totals Source: WDI-World Bank Source: BCRP, SUNAT and World Bank 6. Mining revenues are an important source of volatility, even though they account for less than 10 percent of the fiscal revenues. Between 2001 and 2014, mining revenues (including contributions and levies) accounted for an average of 7.3 percent of tax revenues, below the share of this sector in the gross value-added, which amounted to 10 percent on average. By type of tax, mining accounts for a high percentage of the corporate income tax (almost 25 percent on average 37 in this period), while its share in the general sales tax (IGV) collection is considerably lower (4.2 percent). However, mining revenues have a significant impact on the volatility of fiscal revenues and account for more than half of the increase of current expenditure between 2004 and 2012, and almost 60 percent of the volatility thereof (see Figure 1-2). 7 This means that just as it can give rise to sudden and significant revenue increases, it may as well cause sudden drops impacting the fiscal balance. This seems to reflect the fact that the evolution of the mining sector has both direct and indirect effects on the economy. Therefore, the volatile revenues also appear in the fiscal revenue collection from other economic sectors, e.g. services. 7. Expenditure followed the upward trend of revenues until 2012 with a delay. Fiscal revenues displayed an upward trend since 2005 while expenditures started to go up between 2007 and 2008 (see Figure 1-3). Nonetheless, spending continued to go up during the global financial crisis in 2008-09 but it was back under control in 2011. Once revenues recovered and went back to record levels, spending once again gained momentum as of 2012-13. This new phase of increased spending lasted until 2015 and was closely linked to the public-sector reforms that led to more spending on wages, goods and services (see Figure 1-4). Figure 1-3 Evolution of fiscal revenues Figure 1-4 Evolution of public spending and spending, 2004-2016 2006-2016 Source: BCRP Source: BCRP 8. The overall fiscal balance was in a comfortable position until 2014. In six out of the last twelve years, Peru has had a positive fiscal balance. Between 2005 and 2016, the average fiscal balance amounted to 0.5 percent of GDP, with a minimum of -2.7 percent in 2016. In this period, the ups and downs of the fiscal balance were primarily linked to volatility on the revenue side. However, it is also clear that from 2008, and more emphatically from 2012, the fiscal result started to be disconnected from the revenues due to the increase in spending. 9. The public surplus resulted in a significant public debt reduction . Between 2005 and 2012, the difference between the growth of revenues and expenditures, which increased at a rhythm closer to that of GDP, allowed for significant fiscal savings. This materialized in positive fiscal 7 Between 2004 and 2012, the annual differences in taxes and other contributions collected from the mining sector explain, on average, around 60 percent of the annual differences of the general government’s current revenues. 38 results, averaging 1.3 percent of GDP, to some extent induced by the favorable international context. In turn, this was translated in a significant public debt decrease, which fell from 45 percent of GDP in 2004 to 20 percent in 2012. The sovereign risk decreased significantly, which also helped enable a significant decrease of the financial costs of the public and private sectors (see Figure 1-5). Fiscal savings were also a factor enabling exchange management, the accumulation of reserves and the increase of national savings (see Table 1-1). Table 1-1 Revenues, expenditures and fiscal balance 2005-09 2010 2011 2012 2013 2014 2015 2016 I. General Government Revenues 20.9 21.0 21.8 22.4 22.3 22.3 20.1 18.6 A. Current Revenues 20.3 20.8 21.7 22.4 22.1 22.2 20.0 18.5 1. Tax Revenues 15.5 15.7 16.4 17.0 16.8 17.0 15.2 14.0 Personal Income Tax 1.5 1.6 1.8 1.9 2.0 2.0 1.8 1.7 Corporate Income Tax 4.3 4.6 5.4 5.4 4.7 5.0 3.8 3.4 Sales Tax (IGV) 7.8 8.5 8.6 8.7 8.8 8.7 8.4 8.0 Selective Tax on Consumption (ISC) 1.4 1.1 1.0 1.0 1.0 0.9 0.9 0.9 Others - National (incl. devolutions) 0.2 -0.4 -0.7 -0.4 -0.1 -0.1 -0.3 -1.2 Tax Rev. Local Governments * 0.3 0.4 0.4 0.4 0.4 0.4 0.4 0.4 2. Non-Tax Revenues ** 4.8 5.0 5.3 5.4 5.3 5.2 4.8 4.5 B. Capital Revenues 0.1 0.2 0.1 0.1 0.2 0.1 0.1 0.1 II. General Government Spending 18.0 21.0 19.8 20.4 21.6 22.6 22.3 21.2 A. Current Spending 14.2 13.6 13.4 13.6 14.3 15.5 15.8 15.3 Wages 5.3 5.0 4.9 5.0 5.4 5.9 5.8 5.9 Goods and Services 4.8 5.3 5.1 5.6 5.8 6.1 6.6 6.2 Transfers 3.8 3.3 3.5 3.0 4.1 3.5 3.4 3.2 B. Capital Spending 3.8 6.2 5.2 5.7 6.1 6.0 5.4 4.7 C. Interests 1.8 1.2 1.2 1.1 1.1 1.1 1.0 1.2 III. Non-Financial State Corporations 0.2 -0.2 0.0 0.2 0.1 -0.1 0.1 -0.1 IV. Economic Result (I-II+III) 0.9 -0.2 2.0 2.3 0.9 -0.3 -2.1 -2.7 Memo 1: Revenues from Mining and Hydrocarbons 2.7 2.7 3.4 3.3 2.4 2.3 1.3 1.0 Memo 2: Public Debt 31.9 24.4 22.3 20.8 20.0 20.0 23.3 23.0 * Includes property taxes. ** Includes social contributions, as well as royalties and other contributions from the Mining and Hydrocarbons sectors. Source: BCRP, SUNAT and MEF. 39 Figure 1-5 Public debt and sovereign risk, Figure 1-6 Changes in the fiscal policy annual average stance, 1996-2015 Source: BCRP Source: BCRP, MEF Note: The difference in the primary result is used as a proxy for the fiscal policy stance. 10. Overall, Peru’s fiscal policy tended to be countercyclical. Thanks to the comfortable fiscal position and the low level of public debt, Peru had the necessary space to implement slightly expansive fiscal policies in complicated times for the economy (see Figure 1-6; Box 1-1). Two factors contributed to the ample macro-fiscal space: i) the fast acceleration of fiscal revenues in the past decade, and ii) the authorities’ willingness to follow the fiscal discipline guidelines in effect since the 1990s. Spending tended to be slightly countercyclical despite the absence of fiscal stabilization mechanisms on this side (see box 1-1). 40 Box 1-1 Peru: Fiscal stabilizers Faced with a recession (boom), fiscal stabilizers allow for a reduction (increase) of the fiscal revenues and an increase (decrease) of the expenditures, increasing (lowering) the deficit. This helps soften a recession (boom) and reduces the volatility of GDP. The habitual stabilizer on the revenue side is the income tax and, on the expenditure side, the unemployment insurance and transfers for social programs. A good indicator of the fiscal stabilizers is the cyclical fiscal balance, which captures the cyclical component of fiscal revenues and expenditures. While revenue stabilizers are available in Peru, the ones on the expenditure side are virtually non-existent . Figure B.1.7 shows that, although relatively small, stabilizers on the revenue side are present in Peru. On the other hand, the cyclical component of spending is quite low, suggesting the inexistence of stabilizers on this side. This is consistent with both the inexistence of publicly-financed unemployment insurance and the slow variation of beneficiaries of the social programs. In this context, the anti-cyclical movements of the expenditures are the result of government policies, but not of the automatic change in any spending variable. There is a trade-off between automatic stabilization and spending flexibility. The stabilizers on the expenditure side act through rigid expenditures. On the one hand, the evolution of these components will have an anti-cyclical effect and, on the other hand, this may be an obstacle given the need to adjust spending. Since the beginning of this decade, Peru and other emerging countries have faced a declining commodities price (structural shock). This change entails the need for adapting to the new lower-revenue situation in view of its permanent nature. In this context, the government has to adjust its expenditure account, which is easier given absent factors imposing rigidities on spending. Figure B.1-7 Cyclical balance (proxy of fiscal stabilizers), average of 2005-2014 Source: MFMod - World Bank Recent dynamics on the revenue side 11. Fiscal revenues decreased significantly between 2012 and 2016 (see Figure 1-8). From 2013, the fiscal revenues started to grow below GDP as a result of the declining export prices, particularly copper, and the slowdown of GDP. The decline was even more severe in 2015 and 2016 as a result of the tax reform introduced at the end of 2014, which comprised reductions of the corporate and employee income tax. Regarding the corporate income tax, the reform entailed a gradual reduction of the rate, from the 30 percent in effect in 2014 to 26 percent as of 2019. In 41 the first phase, between 2015 and 2016, the tax rate was lowered to 28 percent. 8 Regarding the employee income tax, the three-tier structure was replaced by a new five-tier structure, lowering the effective rate for most taxpayers. Estimates are that the 2014 tax reform reduced tax revenues by 1 percent of GDP between 2015 and 2016. 9 Figure 1-8 Declining revenues between 2012 Figure 1-9 Declining revenues between and 2016, by economic cause 2012 and 2016, by type of tax Source: BCRP, MEF, WB estimates. Source: BCRP, SUNAT, WB estimates 12. The low tax base, the many tax exemptions, tax evasion, and the low collection efficiency have limited tax revenue collection. Personal income taxes are collected from a small group of mainly salaried workers. A high eligibility threshold effectively exempts almost 80 percent of the registered taxpayers and the rates for the highest quintile are relatively low. In addition, revenues from capital are generally subject to lower rates compared to revenues from labor. On the other hand, collection of the value-added tax is less efficient than comparable countries in the region. Tax exemptions account for around 2 percent of GDP and chiefly apply on indirect taxes. Collection from other taxes, e.g. the general sales tax (IGV), the selective consumption tax (ISC) and the real estate tax, is also much lower than their potential. Corporate income tax evasion amounts to around 50 percent of potential collection according to information provided by SUNAT. Tax evasion and informality are both high in Peru. 8 This reform has been practically reversed through Legislative Decree 1261, which increased the corporate income tax rate to 29.5 percent as of 2017. 9 Revised Multi-annual Macroeconomic Framework (MMMR) 2017-2019. 42 Figure 1-10 Tax revenues in Peru and peer countries, 2014 Source: OECD. Note: Does not include social security contributions. ‘Goods and services’ includes the value-added tax and the sales tax. 13. On average, tax rates are slightly lower than in peer countries in the region; while the corporate income tax rate is significantly higher than the OECD average, the indirect tax rates are lower. Corporate income tax collection is highly concentrated in a reduced tax base, partly due to informality—which is typically affected more by the direct taxes imposed on the capital and labor factors.10 In 2016, 78 percent of the revenues collected from this tax came from less than 13 thousand companies. Concerning indirect taxes, the value-added tax rate is lower compared to countries like Colombia and Chile, and also lower than the OECD average. At the same time, the productivity in collection is substantially lower than in comparable countries (see section 5.3). 10 For example, the 2012 tax reform in Colombia which aimed to reduce informality, was based on the reduction of non-wage costs paid by employers (health fund and social security contributions, and training), which was financed with an increase of the corporate income tax and other consumption taxes. Osorio-Copete (2016) estimates that, as a result of this reform, informality decreased by more than two percentage points. On the other hand, Brazil also implemented a tax reform in 2012 in some sectors of the economy, which consisted on a reduction of the contribution to the pension system, financed with an increase of the gross income tax. Scherer (2014) points out that formal employment grew by 14 to 16 percent as a result of this reform. 43 Figure 1-11 Corporate income tax rate, by Figure 1-12 Value-added tax rate, by country country Source: Deloitte, KPMG. Note: In 2015 and 2016, Peru Source: Deloitte, KPMG. Note: According to a recent had a rate of 28 percent, which went up to 29.5 percent law, under certain conditions the value-added tax would for 2017. go down to 17 percent in Peru as of July 2017. Recent dynamics and selected topics concerning expenditures Recent dynamics 14. Since 2013, the current expenditures for wages, and goods and services, have grown at a faster pace. Current spending had been growing at an average real rate of 6 percent between 2011 and 2012, to then increase to almost 11 percent in 2013 and also in 2014. This was linked to the implementation of reforms in various sectors that required greater current expenditure commitments. Finally, the real growth of current spending slowed down to 5 percent in 2015 and to 0.5 percent in 2016. On the other hand, capital expenditure first slowed down and then declined, particularly in 2016. The net result is that non-financial expenditure increased only slightly, from 19.6 percent of GDP in 2012 to 20 percent in 2016 (see Figure 1-13). 44 Figure 1-13 Spending increase between 2012 Figure 1-14 Spending increase between and 2016 by economic classification 2012 and 2016 by functional classification Source: BCRP Source: MEF and BCRP 15. The rise in current expenditures for goods and services (including public procurement) and wages in recent years was driven by the education, health and citizen security reforms. Regarding wages, the higher payments in education, health and citizen security explain almost 70 percent of the total increase in this item (see Figure 1-15). It should be noted that more than 90 percent of the increase in the education sector was explained by the higher wages for teachers, more than 80 in the health sector by the higher wages for doctors and nurses, and 100 percent in the order and security sector by the higher wages for the police force. Regarding goods and services, the increasing purchases of the three sectors account for almost half of the total increase. In state procurement, there have been significant increases also in other state functions e.g. economic administration, transport and social protection, which were not related to any specific reforms. 45 Figure 1-15 Main accounts explaining the faster public spending (functional classification, % of GDP) Source: MEF, World Bank. 16. Even though the increased spending over the last four years was central government- led, there was a significant increase in spending in some regions . Although the increase of non- financial expenditure since 2012 is explained almost solely by the increased expenditure on central government wages and goods and services, some regions also increased their spending. Map 1- shows each region’s share in the increase of regional government spending, which is called momentum. The regions with the highest share of spending increase were Piura, Lambayeque, La Libertad, Apurímac and Cuzco; while Tacna, Moquegua, Madre de Dios, Ancash and Pasco had a very low share. 46 Map 1-1 Contribution to the total increase in regional government spending, 2015/2012, percentage points Source: MEF, World Bank. Note: The figures on each region represent the share in the increase of the regions’ total spending. The darker-blue regions are the ones with the greatest share. 17. Peru stands out because of its high level of public investment, particularly by the subnational governments. The levels of public investment are higher than the regional, structural and OECD peers (see Figure 1-16). Nonetheless, this higher expenditure on infrastructure is relatively recent. Between 2001 and 2007, the government’s capital spending accounted for an average of 3 percent of GDP, quite similar to its regional and structural peers. But as part of the response to the global financial crisis of 2008-09 capital spending started to go up, and between 2008 and 2015, the percentage remained at an average of 5.8 percent of GDP. 47 Figure 1-16 Types of public spending, % of GDP, 2015 Source: MEF, World Bank. 18. However, current public spending, particularly wage expenditures, is lower than in peer countries. Peru’s relatively low levels of public revenues and expenditure are clear in its current expenditure. On the one hand, this seems to reflect the low appreciation of the civil service and, on the other hand, a low expenditure on public infrastructure maintenance. With regard to civil service, while wage expenditure in Peru accounts for less than 6 percent of GDP, this item accounts for 7.3 percent among regional peers, 9.1 percent among structural peers and 10.5 percent in the OECD (see Figure 1-16). This despite the reforms in education, health and citizen security that have taken place in this decade, of which one of the pillars was wage increase. 19. Education and citizen security account for a relative high share of the budget, reflecting government’s priorities in public service provision. The budget share in education is 17 percent of the total budget, which is higher compared to the budget of structural peers and the OECD, and identical to peers in the region (see Figure 1-17). However, since the size of the budget is proportionately smaller in Peru, the expenditure on education relative to GDP is lower than that of the structural and regional peers and the OECD. Citizen security is a similar case, in which despite the significant budget allocation, the budget is not particularly high as a percentage of GDP. It should be noted that, just like the health sector, these two sectors benefited from reforms during the previous administration with higher wages for the teachers, policemen and health professionals, as well as with more equipment, pushing expenditure up since 2012. 48 Figure 1-17 Spending by sector, % of total spending, 2015 Source: MEF, World Bank. Selected topics 20. The rigidity of spending, i.e. the flexibility with which spending can be adjusted, has increased. The proportion of rigid expenditures has increased from 66 to 70 percent between 2012 and 2015 (see Figure 1-18). In current expenditure, the proportion of rigid expenditures is quite high and rose from 92 percent in 2012 to 93 percent in 2015 (see Figure 1-19, Table 1-2). This is primarily explained by higher spending in wages over this period. In capital expenditure, the rigidity has increased from 6 to 11 percent, which is linked to the greater payment commitments with the PPPs. 49 Figure 1-18 Spending rigidity, percentage Figure 1-19 Spending rigidity, by type of of total spending expenditure, percentage of category total Source: BOOST, MEF and WB estimates Source: BOOST, MEF and WB estimates Table 1-2 Classification of expenditures by degree of rigidity Staff expenditure (permanent payroll), social security contributions, interests, payment High rigidity commitment (including PPPs) Medium-high Staff expenditure (temporary payroll), current transfers to educational institutions rigidity Medium rigidity Non-staff services, other current transfers Consumption goods, other current expenditures, non-financial direct investment, financial Low rigidity investment, other capital transfers Source: World Bank. 21. Moving forward, it will be important to look into the possibility of enhancing the flexibility of some of the more rigid expenditures. Another possibility is to analyze the expenditures’ rigidity based on an analysis of variance (Barrot and Battaile, 2013). Using the economic classification of expenditures, the matrix of variance and covariance is obtained whereby each component is divided by the total spending variance. According to the results, three spending components have a greater share in the total spending variance: staff, services and non-financial assets. These components also have a high share in the spending. This analysis suggests that in view of their greater volume and variability, these components should also be part of a medium- term fiscal adjustment strategy, though this would require a thorough functional and staff revision in the different government units (see Table 1-3). 50 Table 1-3 Analysis of variance and covariance of public spending, by generic expenditure account Personnel Pensions Goods Services Donations Other Donations Other Procurem Procurem Public and Social and Other and Expenditu and Expenses ent of Non- ent of Debt Obligation Social Transfers res Transfers Financial Financial Service s Benefits Assets Assets Personnel and Social Obligations 8.5 Pensions and Other Social Benefits 0.8 0.1 Goods 1.4 0.1 0.2 Services 7.4 0.7 1.2 6.5 Donations and Transfers 0.8 0.1 0.1 0.7 0.1 Other Expenditures 2.2 0.2 0.3 1.9 0.3 0.7 Donations and Transfers 1.0 0.1 0.2 0.9 0.1 0.3 0.1 Other Expenses -0.1 0.0 0.0 -0.1 0.0 0.0 0.0 0.0 Procurement of Non-Financial Assets 5.3 0.6 1.0 4.7 0.5 1.3 0.5 0.0 4.9 Procurement of Financial Assets 0.2 0.0 0.0 0.2 0.0 0.0 0.0 0.0 0.1 0.0 Public Debt Service 1.4 0.1 0.2 1.2 0.1 0.4 0.2 0.0 0.8 0.0 0.3 Source: MEF and World Bank estimates. Note: variance of series between 2009 and 2016. Box 1-2 Peru: Fiscal rules In 1999, Peru adopted its first version of fiscal rules through enactment of the Fiscal Responsibility Law, which set limits on both the deficit and expenditure. Later, these rules were modified through the Strengthening Responsibility and Fiscal Transparency Law (Law No. 30099) which was passed at the end of 2013. This new framework considered a policy guide based on the structural balance and provided for a limit of 30 percent of GDP for public debt. In December 2016, the Legislative Decree No. 1276 was published establishing a new fiscal framework for the non-financial public sector (NFPS), subject to the following rules:  Rule on NFPS Debt: debt should not exceed 30 percent of GDP, with an exceptional deviation of 4 percentage points in case of financial volatility.  Rule on economic result: the fiscal deficit of the NFPS cannot exceed 1 percent of GDP.  Rule on non-financial expenditure of the general government: real growth cannot be higher than the average of 20 years of the average GDP real growth (15 previous years, the present year and the projection for the coming 4 years) plus 1 percentage point.  Rule on current expenditure for the general government: after excluding the expenditure for maintenance, real growth cannot be higher than the average of 20 years of the average GDP real growth (15 previous years, the present year and the projection for the coming 4 years) minus 1 percentage point. The rule on the economic result is not enforced immediately since the Law No. 30499 of August 2016 defined the rule for the year 2017 at a deficit of 2.5 percent of GDP. Moreover, through a transitory provision it was determined that the fiscal deficits of years 2018, 2019, 2020 and 2021 cannot exceed 2.3, 2.0, 1.5 and 1.0 percent of GDP, respectively. Fiscal Sustainability Analysis Fiscal Consolidation 22. The contraction of fiscal revenues and the pressure of current expenditures were reflected in a deterioration of the fiscal balances. The fiscal balances deteriorated fast and the surplus exceeding 2 percent in 2012 became a deficit of 2.7 percent in 2016. By August of last year, the fiscal balance reached a peak of 3.4 percent of GDP, and the authorities began with the fiscal adjustment, so that the target of 3 percent would be reached by December. Thanks to a 51 combination of an adjustment that was slightly greater than estimated and extraordinary fiscal revenues by the end of the year, the deficit exceeded the target and amounted to 2.7 percent. This helped soften the adjustment henceforward. 23. The fiscal responsibility and transparency framework provides for a gradual reduction of the fiscal deficit to 1 percent in 2021. The government has formulated a trajectory for fiscal adjustment which is described in the MMMR 2017-19 and Legislative Decree 1276 (Table 1-4). In accordance with this adjustment, the fiscal deficit would be reduced gradually until reaching 1.0 percent in 2021. While the projection is that general government revenues will increase to 19.5 percent in 2019, expectations are that the expenditures will remain just slightly above 21 percent by 2019, i.e. the fiscal deficit in that year would be 2 percent of GDP. Thereafter, an additional adjustment of 1 percentage point of GDP would have to be made until 2021 in order to reach the proposed deficit. 24. The passive scenario, which assumes a fiscal deficit of 2.7 percent until 2021, is not compatible with stabilization of the debt ratio. Under this assumption, the debt ratio would go up in a sustained manner and exceed 30 percent of GDP in 2021 (see Figures 1-20 and 1-21). 25. In the baseline scenario, which follows the evolution of the deficit assumed in the MMMR 2017-19, the debt ratio stabilizes. Following the fiscal responsibility framework, and considering the GDP growth assumptions, the exchange rate and the export prices suggested in the MMMR 2017-19, the debt ratio would reach a maximum of 25 percent of GDP in 2019 to then decrease slightly until 2021 (baseline trajectory, Figure 1-21), according to our estimates. Assuming the financial revenues and expenditure remain constant on the level projected for 2019 by the MEF, it would be necessary to have an adjustment of 1 percent in the financial expenditure until 2021 for the fiscal deficit to reach the government’s objective. This expenditure cut is equivalent to a real growth decrease of the non-financial expenditure, from a rate of 6.6 percent in the last ten years to an annual 3.2 percent between 2017 and 2021. Table 1-4 Assumptions of the baseline scenario 2017 2018 2019 2020 2021 GDP (Var. %) 4.8 4.3 4.1 4.0 4.0 Currency depreciation 3.3 2.0 0.0 0.0 0.0 Export prices (Var. %) 2.4 1.4 0.9 0.0 0.0 Interest rate (%) * 5.5 5.4 5.2 4.8 4.8 * Cons i s tent wi th payment of i nterests of the MMMR 2017-19. Source: Revised Multi-Annual Macroeconomic Framework 2017-2019 (MMMR 2017-19) 52 Figure 1-20 Fiscal deficit under various Figure 1-21 Evolution of the public debt scenarios under various scenarios Source: MEF and WB estimates Source: MEF and WB estimates Note: The passive scenario is one in which the level of non-financial expenditure is maintained constant as a fraction of GDP at the level of 2016, whereby the other assumptions come from the MMMR 2017-19. The baseline scenario is one consistent with the projection of the MMMR 2017-2019 for the fiscal accounts and for the other macroeconomic variables (see Tables 1.1 and 1.2). The estimated shocks are: i) an interest rate of 1 pp higher on average in 2017- 2021, ii) constant export prices at the level of 2016, iii) a reduction of 1 pp in the IGV as from 2017, iv) a lower GDP growth of one percentage point per year, v) depreciation of 10 percent in 2017 and 2018 (increases the level of debt but increases the fiscal revenues, and therefore reduces the effort of the adjustment), and vi) combined (the result of which is lower than the sum of the negative shocks because of the buffer effect of a higher exchange rate). 26. In contrast, under the alternative assumption that GDP grows less than pointed out in the MMMR 2017-19, the public debt would continue to increase, approaching the upper debt limit of 30 percent. A GDP growth of one percentage point lower than mentioned in the MMMR 2017-2019 would lead to a lower tax collection and cause a significant deviation of the fiscal deficit and increase the debt ratio, bringing it nearer the limit of 30 percent in 2021. Moreover, in a stress scenario in which a slower growth is combined with export prices that do not increase, a higher interest rate, a depreciation of the exchange rate and an IGV tax reform, the deficit and debt would go up faster compared to the passive scenario. 27. Consequently, a greater reduction of the non-financial expenditure is required to ensure a sustainable debt trajectory in the event of adverse shocks . Faced with the shocks set out on Figures 1-20 and 1-21, the government’s revenues would be much lower than expected. Hence, to maintain the path of the deficit and debt, a greater spending adjustment would be required. In these alternative scenarios, the reduction of non-financial expenditure needed for the fiscal balance to follow the path proposed by the government would amount to between 1.4 and 3.0 percentage points of GDP (see Figure 1-22). This would require the slow-down of the non- financial expenditure at an annual average rhythm of between zero and 2.8 percent from 2017 to 2021. 53 Figure 1-22 Adjustment of non-financial expenditure by 2021 under different scenarios Source: World Bank estimates. Note: In the different scenarios, the financial revenues and expenditure vary. The figure only shows the adjustment in public spending needed to reduce the fiscal deficit to 1 percent of GDP in 2021. Contingent risks derived from the PPPs 28. The firm commitment to PPPs is having a growing impact on the budget. Peru is intensively using private investment to build public infrastructure. Between 2004 and 2016, 77 PPP contracts were awarded for a total investment of more than US$25 billion (without IGV), 20 of which were co-financed PPPs. The payment commitment accounted to 0.7 percent of GDP in 2014 and almost 1.7 percent of GDP in 2017, more than one third of the country’s investment budget. 29. The explicit guarantees linked to the PPP contracts were estimated at 2.8 percent of GDP at the end of 2015.11 Nonetheless, inadequate structures, limitations in instruments to assess the risk distribution between the public and private sectors, and the issuance of payment certificates may give rise to excessive non-explicit risks which may materialize in the future. Aside from other issues related to risk distribution, the Government implicitly assumes two risks which the international practice considers should be taken on fully by the private sector: the construction risk (except in complex constructions like tunnels) and the financial risk. Furthermore, the system also increases the probability of the completion risk (i.e. that at a certain point in time the concessionaire leaves without completing the works or leaving behind major liabilities for the Government). 11 According to the Explicit Contingencies Report published by the MEF, at the end of 2015 the maximum exposure of the NFPS to explicit fiscal contingencies amounts to 9.7 percent of GDP. This is derived from judicial claims and arbitration proceedings (5.5 percent), guarantees associated to the execution of PPPs (2.8 percent), international investment disputes (0.8 percent) and guarantees or sureties (0.5 percent). 54 30. The contingent liabilities have started to materialize in terms of constant addendums that almost always contain additional commitments and contingent elements. Even though a certain degree of adjustment to the long-term contracts is reasonable, most OECD countries do not permit renegotiations with an increase of more than 20 percent in terms of the time or cost of the service. An Arup study that evaluated two PPPs suggests that a large part of the contingencies associated to one of these projects is not identified and the initial budget submitted may easily double (see Figure 1-23).12 This underscores the need for making more detailed calculations of the contingencies to evaluate the fiscal impact thereof. Figure 1-23 Simulation of the fiscal impact of the contingencies of the Chavimochic project Source: MEF and Arup 31. Introducing a new fiscal framework for the PPPs would maximize the fiscal space for investment and limit the fiscal risk. The advantage of using a PPP scheme for public works is a twofold: i) it reduces the impact of the infrastructure investments in the budget, and ii) it encourages development of the capital market. The present impact of the PPPs on the multi-annual macroeconomic and fiscal framework is very similar to that of public works. In order for the benefits of PPPs to materialize, the recommendation is to introduce a clear fiscal framework that has the following elements:  Clear and permanent multi-annual ceilings. For example, Colombia limits the space for PPP co-financing to an annual maximum of 0.4 percent of GDP, which has led to the obligation of the works being structured in such a manner that they attract financing at 15 to 25 years. 12 Arup (2016), Mejora del Mercado de APP en el Perú, Revisión Ex Post de Proyectos APP, March 2016, mimeo. 55  Introduce a clear risk allocation system with savings implications (provisioning) for the awarding entities and with specific records in the fiscal accounts.  Improvements in future resource allocation processes in order to generate credibility and avoid the need for issuing certificates. One example in this sense is the future budgets in Colombia.  Assess the capacity of the user to act as an additional source of PPP financing. 32. Furthermore, the MEF and Proinversión could lead a change in the PPP structuring process. This change could entail the following elements:  A model contract for PPPs with performance indicators and payments that reflect the required service or functionality of the asset rather than progress of the works.  Restructure risk allocation so that all construction, financing and other insurable risks remain with the private party, and so that the property, environmental and network risks would be shared.  Significantly improve the pre-investment phase, so that the traffic, engineering, geological, environmental and property studies would be of a high quality.  Pre-selection of potential concessionaires (focusing on their structuring and financing capacity rather than on their construction capacity) and elimination of very low bids.  Development of enhanced financial products by COFIDE or any other similar entity in order to attract more sources of financing and improve terms and rates. Options to implement fiscal consolidation 33. This study identifies sectoral spending reforms that may lead to savings of at least 1 percent of GDP through efficiency improvements that do not affect equity. Moreover, it suggests additional reforms that would rise spending efficiency but are harder to quantify in terms of short-term fiscal savings. These measures are identified in the fields of the formulation of public procurement, citizen security, water and sanitation, transport, health, education, risks and budget. With these savings, it will be possible to obtain the same level of services with fewer resources or, alternatively, achieve an expansion of the services with the allocated resources. In the latter case, the savings would soften the expected slowdown effects of public spending. Since additional fiscal space would have to be generated in case of any of the unexpected stress scenarios materializing in macroeconomic variables, this study also identifies other efficiency-enhancing reforms, which are more difficult to quantify in the short term in terms of fiscal savings. The specific reforms are summarized in Table 1-5, in addition to the direction in efficiency and the fiscal savings of every reform. Possibly, this table would identify the general direction of the impact of the measure on spending equity. 56 Table 1-5 Overview of policy options and their impact on efficiency, equity, and fiscal savings Potential Savings (% Sector Measure Efficiency Equity of GDP) Reform of the procurement system, with the following options in terms of intensity: + Neutral - Full reform of the procurement system (aggressive approach) * + Neutral Public Procurement - Full reform of the procurement system (conservative approach) * + Neutral 0.4% - Reform of 9 low-value and low-risk categories (aggressive approach) * + Neutral - Reform of 9 low-value and low-risk categories (conservative approach) * + Neutral Optimization of human and physical resources at the district level + + 0.3% Citizen Security Creation of the resource allocation formula with efficiency measures + + TBD Funding of programs for the construction of police infrastructure + Neutral Reduce the undervaluation of water and sanitation services for EPS + 0.1% Improve the EPS' collection efficiency + 0.0% Reduce the losses of unaccounted-for water in the areas in which the EPS operate + 0.0% Improve planning and coordination of the total investment and the investment by + + 0.1%-0.2% government level Water and Sanitation Evaluate the specific viability of every project as part of a greater investment program, which may help reduce the transaction costs for project preparation and + + implementation TBD Improve the structure incentive in the budget process: (i) decoupling the budget process and investment to enable multiannual planning of the investment and budget, + and (ii) changing the rules that create adverse incentives. Multiannual planning of transport works + + Transport Implement the multiannual budget with MTC as the unit coordinating different levels of TBD + Neutral government Continue to expand the population coverage of insurance Neutral + -- Redefine the package of benefits, keeping in mind the fiscal impact and the impact in + + terms of equity Adapt international payment systems for suppliers, incentivizing the efficient + Neutral production of services Improve efficiency in the purchase and distribution of medicines + Neutral TBD Health + Neutral Build the institutional and operational foundations for the interoperability of insurance Reconfigure specialized hospital care programs + Neutral Unify complementary services at the regional level + Neutral Lower the administrative expenses of the health system + Neutral 0.1%-0.2% Remove the implicit subsidies for EsSalud + Neutral 0.1% Provide adequate financing of the current expenses for the provision and supervision + + of educational services in different contexts, integrating it into results-based budgeting Allocate the investment budget in a multiannual manner, according to the priorities and strategies of the National Plan for Educational Infrastructure (PNIE), budgeting also the + Neutral necessary operating expenses Prioritize the expenses in highly effective and pro-poor interventions and policies + + Education Implement an electronic government that connects schools and optimizes information + + TBD flows, reduces transaction costs and speeds up administrative procedures Establish differentiated strategies (by region, university) and accredit capacities at the + + level of schools and education units in order to enhance budget responsibilities Establish the conditions for allocating the subnational budget by means of student- based formulas that comprise equity criteria, starting with accredited schools and + + education units. Link the budget to performance-driven incentives. + Neutral Contingencies: explicitly allocate public and private risks in new PPP contracts, as well Risks + Neutral TBD as the instruments for mitigation Budget: reduce variability of the PIA versus the PIM, allocating a greater proportion of Budget + Neutral TBD the budget at the beginning of the year TOTAL 1.0%-1.2% * The conservative approach merely requires improvement of the existing procurement strategies. The aggressive approach requires reforms of the conservative approach plus an optimization of the levels of consumption, which may require a reduction of current levels or the replacement of products. Source: World Bank. 57 34. Beyond, it is possible to consider measures on the revenue side to create additional fiscal space to counteract potentially less benign macroeconomic scenarios or to address the growing pressure for quality public services. Tax collection increased from 14 percent of GDP in 2004 to 17 percent in 2014. However, the 2014 fiscal reform and the lower corporate earnings have lowered this once again to 14 percent of GDP by the end of 2016. This is low relative to some of Peru’s regional and structural peers. Moreover, with this level it is not possible to satisfy the demands for infrastructure and human capital accumulation within a reasonably short period. 35. The low productivity of the IGV suggests that it may be possible to increase collection by increasing the tax base and reducing evasion. Productivity of the value-added tax amounts to only 60 percent in Peru, i.e. lower than the average of the OECD and various countries in Latin America (see Figure 1-24). This is related to both evasion and erosion of the tax base with the existing exemptions. It should be noted that some of these exemptions did not achieve their original objective, which was to benefit the low-income population. On the contrary, the exemptions encouraged evasion and in some cases promoted illicit activities, with an impact on productivity of the tax. In tax policy terms, the tax expenditure total cost on the IGV (mainly as a result of the exemptions) amounted to 1.6 percent of GDP in 2014. An in-depth analysis of the effectiveness of these exemptions might provide clear guidelines about which ones could be eliminated with a low social and economic cost and a higher fiscal gain. In terms of administration, it is also necessary to combat IGV evasion by means of strategies to simplify and incentivize the use of payment vouchers. A reduction of the evasion of this tax from 30 percent to 20 percent would increase collection by almost one percentage point of GDP. Figure 1-24 IGV Productivity13 Source: OECD, KPMG, WDI, World Bank calculations. 13 ⁄ IGV productivity is measured through C-efficiency, which is calculated as where R is IGV collection, C is private consumption and  is the tax rate. This calculation is based on the methodology used by Gómez and Steiner (2015). 58 36. A more intensive use of the ISC may enhance its effectiveness to discourage activities that generate externalities, while at the same time increasing collection . ISC collection is under one percent of GDP, significantly lower compared to the region (1.8 percent) and the OECD (2.7 percent). The adoption of international best practices requires product-level studies to determine the optimal level. This would be the basis to redefine the fuel, alcohol and cigarette tax bases and rates. Moreover, the imposition of green taxes on dangerous gases that are harmful to health or greenhouse gases should be considered, following the recent example of Chile. In this context, it may be worthwhile to also consider a redefinition of the tax on vehicles up to 3 years old versus more efficient schemes that tax vehicles with the greatest fuel consumption. 37. An improvement of the control and audit systems would help reduce income tax evasion. Evasion of the corporate income tax is high, above 60 percent of the potential collection. Although the direct impact of the audits on collection is low, the primary objective is not to impose sanctions, but to encourage voluntary compliance through a greater perception of tax evasion and tax avoidance risks. Furthermore, Peru should align to international best practices to combat the avoidance and transfer of profits (BEPS) with a focus on multinationals. For this purpose, the anti- avoidance clause enacted in 2012 (Legislative Decree 1121) should be implemented gradually and prudently, which also requires strengthening of the audit function and the tax administration capacities. 38. The reduction of the labor costs and the simplification of the income tax structure would enhance formalization. Recent World Bank estimates suggest that personal income tax evasion accounts for 36 percent of the potential collection, half of which is linked to undervaluation in the tax return. A reduction of evasion by 10 percentage points would be equivalent to an additional tax collection of 0.2 points of GDP. Simplifying the structure, the tax return and payment would be an incentive to reduce evasion. On the other hand, the comparison to regional peers suggests that the fringe benefits (compensation for time of service, bonuses and others) are quite high in Peru, while their effective collection is low. The multiple labor costs and taxes, as well as the high marginal rates, foment informality and evasion. In Brazil and Colombia, for example, the tax reforms that reduced the non-wage costs enhanced the formalization of companies and employment. Colombia lowered the contributions for the health fund, social security and training for workers, financing this with a corporate income tax.14 These countries experiences and other Eastern European countries suggest that labor taxes are the most critical taxes in terms of a comprehensive plan to reduce informality. 39. A greater role of the property tax would improve collection of the local governments and may reduce their dependence on central government transfers. Giving greater importance to property tax is a measure that would have more effects in the long term. The first steps to implement this strategy would be to move forward in the formalization of property deeds and increase the number of properties registered in a cadaster. This would reinforce the local governments’ budgetary autonomy and enhance the demand for accountability. 14 According to measurements of Rocha, Ulyssea and Rachter (2016), the reduction of non-wage costs for micro enterprises in Brazil in 2009 increased the proportion of formal enterprises by 4.8 percentage points. Also according to measurements of Morales (2016) and Steiner (2016), the reduction of non-wage costs and the increase in income (total neutral tax reform) also enhanced the formalization of enterprises and employment in Colombia. 59 Figure 1-25 Property tax relative to GDP 1.2 1 0.8 0.6 0.4 0.2 0 2008 2010 2013 Source: OECD, World Bank Aspects of multi-annual expenditure programming that would facilitate fiscal consolidation and improve budget transparency 40. The approved budget does not always reflect the true level of public spending, particularly at the subnational level. The opening institutional budget (PIA) is merely a reference because throughout the year there are a series of additions that lead to the modified institutional budget (PIM), which is the one setting the true limits on spending. At the general government’s level, the 2015 PIM was 17 percent higher than the PIA; nonetheless, at the local governments’ level, the PIM almost doubled the PIA (see Figure 1-26). In the first case, the difference is largely explained by the unused balance from previous years, while at the local governments’ level an important part is explained by the transfers from the central government throughout the year. Figure 1-26 Public budget and expenditure Figure 1-27 Quarterly expenditure, by executed by function, 2015 government level (% of GDP), 2015 Source: MEF and WB estimates Source: MEF and WB estimates 41. The budget rescheduling throughout the year generates political incentives for a competition for the resources. In a context of budget inertia, the two most important criteria for 60 the allocation of expenditures tend to be the past budget and the ratio of execution, which generate incentives sequentially. Due to their nearness to the budget process, the units of the Executive (ministries) have greater bargaining power and first maximize the functional budget. Subsequently, part of this budget benefits local units through transfers. In some cases, the ministries lay down rules for the allocation, while in other cases this is based on a political negotiation. Finally, there is competition in execution, both in the sectoral sphere and by level of government. With regard to the latter, the ministries as the operational arm of the central government have an incentive to maximize their transfers to the local governments, so that their execution ratios would be high, transferring the apparent inefficiency in execution to the subnational governments. The result is that the allocation of resources in the current budget process is not very transparent and gives rise to an administrative burden for the central government bodies. 42. The budget process is highly seasonal, which may be a source of both a rigidity in the expenditure response and of the low levels of execution. In 2015, 62 percent of the local government resources for capital expenditure were budgeted throughout the year. This gave rise to delays in preparation and execution of the projects and, for example, in 2015 almost 70 percent of the capital expenditure was executed in the second half of the year. Furthermore, the levels of budget execution are affected as well. 43. The shortcomings in management of the public investment, combined with the low intergovernmental coordination, undermine the efficiency of spending in public infrastructure. Peru ranked 115 out of 138 countries in the quality of its general infrastructure according to the World Economic Forum report of 2016-2017. The low efficiency of public spending is related to shortcomings in management of the public investments, such as the absence of multi-annual budgets or the deficient national or sectoral planning (Figure 1-28). Moreover, Peru has a low score in the coordination of infrastructure between the central and local governments, which points to decentralization deficiencies. 61 Figure 1-28 Deficiencies in management of public investment Source: IMF (2016). Note: IMF calculations based on surveys in Latin American countries (Argentina, Bahamas, Belize, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, Grenada, Guatemala, Honduras, Jamaica, Mexico, Panama, Paraguay, Peru and Saint Lucia) and 12 emerging countries from other regions. 44. Expensive administrative measures have been implemented, aiming to correct the poor quality of public spending. The high administrative costs associated to the management of public investments are, to a certain extent, the consequence of the decentralized system with small municipalities that generally lack capacity and that execute most public investments. Instead of adjusting the current mechanisms for fiscal decentralization, the government has followed a second-best strategy to impose several administrative quality control mechanisms that in the end increase the costs for administration and coordination of public investments. 45. More efficient capital spending requires a more comprehensive framework for investment planning. Peru uses a standard methodology for the selection and evaluation of projects, but does not publish projections of capital expenditure beyond the present fiscal year. The result is that there are no multi-annual targets or ceilings on the capital expenditure by ministry or program. In addition, there is no official record of the obligations in future years within the framework of the signed public investment agreements. 46. The modified budget should perform its cash management role for the intra-annual funds use optimization. The PIM should not be used to increase the public entities’ resources, as is in the present, but to redistribute resources and for the budget authority to ensure an efficient cash management, transferring resources that will not be used to where they are needed. For this purpose, the budget authority should have the prerogative of making these transfers. Moreover, a ceiling could be implemented on the aggregated budget increases, which would reduce the aggregated variability of public funds. 47. Multi-annual expenditure programming would ensure a greater budget predictability. Expenditure programming would be helpful to align the fiscal objectives to the sectoral priorities defined by the government. Multi-annual programming would help achieve this 62 purpose since it would contain the sectoral expenditure ceilings and commitments for the coming years. This would help align the sectoral objectives to the macro-fiscal strategy. In addition, this would show the direction of the changes and the commitment to fiscal discipline. With no multi- annual programing, there is a risk of the changes in the expenditure being made from a short-term perspective, which would bias the adjustment towards activities that may be reduced in the very short term, like capital spending. Finally, the multi-annual programming should be consistent with the multi-annual macroeconomic framework (MMM), the projections of which should reflect a realistic analysis of the economic variables in order to maintain credibility of the programming. 48. A reform that reinforces the redistributive criteria in the allocation of subnational transfers would enhance the efficiency of all public services . The significant gap between subnational governments own resources and expenditures make them dependent on intergovernmental transfers as a source of financing. While the transfers of ordinary resources are the most important ones for the regional governments, the canon (the system of transfers related to fiscal revenues from the natural resources) has had a special significance among the local governments. The allocation of ordinary resources is not based on criteria such as the unsatisfied demand for public services or the fiscal capacity, but distribution is rather based on supply criteria, such as staff and facilities transferred to the regional governments, or in negotiations. On the other hand, the canon is a compensatory transfer based on geographical origin. It is therefore not linked to the need for spending or the fiscal capacity. This means it has a strong regressive effective, whereby the wealthiest municipalities get more resources. A comprehensive transfers reform would give rise to more equity, with the poorer areas and the areas with a higher unsatisfied demand for services being given priority. A simulation of the transfer of ordinary resources, following the FONCOR criteria (demand-based rules) show that the regions with the highest poverty indices would benefit from this change. In this exercise, the ordinary resources are distributed by using the existing FONCOR formula. The benefited regions would be ones with the highest poverty rates, such as Huancavelica, Apurímac, Pasco, Huánuco, Amazonas and Puno. 15 Conclusions Over the last decade, the Peruvian economy has benefited from a prudent fiscal management. Thus, until 2014, the fiscal balance was in a comfortable position, which led to an important reduction of the public debt and the sovereign risk. Furthermore, fiscal savings allowed the implementation of countercyclical fiscal policies despite the absence of stabilizers on the expenditure side. However, since 2013, fiscal revenues have declined while expenditure increased, leading to a public sector deficit of 2.7 percent of GDP in 2016. If this trend continues, in some stress scenarios the public debt ratio may exceed the legal limit of 30 percent of GDP laid down in the fiscal rule in three years. At the same time, it is important to preserve the fiscal capacity in order to be able to react in a countercyclical manner to negative shocks on the product. Considering various scenarios, we estimate that the adjustment needed to stabilize the public debt ratio amounts to between 1 and 3 percentage points of GDP. This document identifies measures to enhance the spending efficiency that may lead to fiscal savings of at least one percentage point of GDP. Other measures have been identified as well without measuring the effect on the public budget. Moreover, this document presents tax revenue options to complement the fiscal adjustment and 15 The analysis and recommendations are based on the World Bank note of 2017, “Perú: Construyendo un sistema de descentralización fiscal más eficiente y equitativo”. 63 makes recommendations related to the budget process and the expenditure program in order to ensure a greater predictability and transparency, which would also enable the implementation of the fiscal consolidation. 64 Introduction 1. Modernizing Peru’s public procurement strategy has a high potential to bring significant efficiency gains, fiscal savings, and more accountable public resource management. The public procurement of goods, works and services carried out by the general government of Peru add up to more than thirty billion soles in 2014. This represents almost 38 percent of the total general government expenditure or 5 percent of GDP, and thus constitutes a large opportunity to seek for improvements in the efficiency of the services provided by the government through lower prices and better quality. Public procurement in Peru had initially been managed only by the central government, but has become more fragmented through the process of decentralization. Today, it is carried out at all different levels of government, bearing a high risk for efficiency losses through fragmentation, fewer bidders, and less competitive methods used. 2. The analysis undertaken by this review shows that more efficient procurement strategies would yield up to 0.4 percentage points of GDP in fiscal savings. Achieving these savings requires a comprehensive reform of the procurement system for all product categories purchased by the general government of Peru. This chapter considers two reform scenarios. Under the conservative scenario, the implemented procurement strategies focus on leveraging the purchasing power of the government, introducing quality standards to attain lower prices, or increasing the use of more competitive methods. Under the more aggressive scenario, strategy changes also include the adoption of optimization criteria to reduce the current consumption levels and substitute the use of expensive with less costly but qualitative equivalent inputs. The fiscal savings opportunities amount to between 1.1 billion Soles (0.2 percent of GDP) under the conservative scenario and 2.2 billion Soles (0.4 percent of GDP) under the more aggressive procurement strategy scenario. The analysis also shows that small adjustments in strategies, which can be implemented without any modifications to current procurement laws or systems, for only nine standardized goods would yield up to 257 million Soles in fiscal savings each year. A detailed strategic sourcing analysis is then provided for three of these goods, vehicles, cement, and fuel; it reveals a large fragmentation of procurement processes and methods, leading to large dispersion of prices for the same good paid by different government bodies—sometimes even purchased from the same supplier. For instance, the governments purchased 638 sedan cars in 2015 and prices paid varied by 600 percent while almost three-quarter of the procurement processes had only 1 bidder. 3. The chapter is organized in four parts. The first reviews the legal framework and the share of public spending going through public procurement processes. The second part presents the strategic sourcing methodology that is applied to identify more efficient procurement strategies that maximize fiscal savings without compromising the quality of public expenditures. The third part describes the fiscal savings from implementing more efficient procurement strategies. The fourth part provides the policy options needed to realize the fiscal savings. The scope and framework of public procurement in Peru 4. The public procurement of goods, works and services carried out by the general government of Peru add up to 30.31 billion Soles in 2014. This represents almost thirty-eight 65 percent of the total government expenditure or 5 percent of GDP (Figure 2-1). Thus, it constitutes a large opportunity to improve efficiency in the services provided by the government through lower prices and better quality. In Peru, public procurement is carried out at different levels of government (central, departments, provinces and districts), with a scheme that was initially centralized but has applied a process of decentralization during the last years. Figure 2-1 Public budget execution 2014 (in million soles) Source: MEF data; WB staff calculations. 5. The public procurement system in Peru has recently been reformed by the adoption of a new public procurement law and its related regulation.16 The general objective of the reform is to establish norms to maximize value-for-money of invested public resources and to promote results-based management in the area of public procurement. The new framework also aims at ensuring good management in terms of purchasing by public officials, taking into consideration efficiency, transparency and oversight, which represent the basic principles of the public procurement system. The specific objectives of the new law are: (i) to simplify public procurement processes; (ii) to implement Peru Compras, a Centralized Purchasing Agency; (iii) to standardize public procurement processes (through the development of the new tools by Peru Compras); (iv) to increase competition for public contracts; and (v) to incorporate into the legal framework commitments made through international agreements. 17 6. Public entities implement procurement using different methods based on monetary thresholds and in some cases specific justification for direct awards without competition. Public bidding or public tendering as well as reverse auction are the most competitive methods. There are other methods available such as simplified procurement, shopping or direct contracting that are less competitive and should be used in specific circumstances or when the amount is low. 7. The reform provides Peru with a unique opportunity to implement incremental measures to improve procurement results. Based on evidence, this analysis identifies areas that can generate savings by studying the main categories in which public procurement is carried out in Peru, their main characteristics and identifying areas of improvement to implement short term procurement strategies. 16 The New Law N° 30225 – Public Procurement Law became effective on January 9, 2016. It is, available at http://www.osce.gob.pe, 17 “Public Procurement Review of Peru”. Meeting of the Working Party of the Leading Practitioners on Public Procurement (LPP). OECD. October 2016. 66 Strategic sourcing methodology Supply positioning to identify products with high potential savings 8. A “supply positioning” tool, which analyzes all procurement carried out in 2015, was used to identify product families with a high potential for fiscal savings. This analysis allowed a detailed assessment of the expenditure and an estimation of the complexity of the goods, works and services procured by the Government. It provides the possibility of classifying goods, works and services based on three variables: amount of the expenditure, complexity of the market, and number of buyers. The result is reflected in Figure 2-2 and shows in the bottom right corner the type of procurement that represents high volume of expenditure but low risk. These are usually commodities or standardized items and tend to have competitive markets; e.g., fuel, office maintenance, security services, among others. Figure 2-2 Spending analysis of the Central Government (2014 – Boston Matrix, Level 6) Source: MEF data; WB staff calculations. Also Available as an interactive dashboard at: https://tab.worldbank.org/t/WBG/views/PeruPER/3_Spendanalysis?%3Aembed=y&%3AshowShareOptions=true &%3Adisplay_count=no#3 9. The application of strategic procurement policies should start classifying categories according to value and complexity/risk. The development of a proper strategy begins by identifying exactly which categories fit these characteristics of high value and low complexity/low risk, which ones are mostly bought by the government and which ones should have a specific 67 strategy to address them. It is an ideal starting point to bring about change and improvement to the whole system. For example, products with low levels of complexity, are easier to standardize and to procure through centralized purchases (see Figure 2-2 for a classification). Strategic sourcing analysis 10. This study was carried out with the objective of identifying a first stage of strategies that could be implemented in the short term. Three out of the ten families were selected to be analyzed in detail: fuel, vehicles and cement. The identification of these specific items was based on the following characteristics:  They consist of a small group of homogeneous goods and services;  They are easy to standardize with known and established specifications from the market;  They are procured by multiple government entities in a fragmented and dispersed way;  Demand consolidation is simple to increase volume and reduce prices. 11. The methodology to study each of the selected families of products is based on the "strategic sourcing" model. This approach has been chosen by a large number of high performance private and public sector entities for supply and basically consists of the following six milestones with define products. It consists of six steps (Figure 2-3). Figure 2-3 Strategic Sourcing Model Communic Suppliers / Spend Market Strategy for ation Bidding Implementation analysis analysis savings program documents Source: World Bank. 12. The first step is the spend analysis and consists of a review of the current procurement arrangements and the results of implementing them. This includes an understanding of the procurement processes, identification of internal and external restrictions or limitations on each of the selected procurement item groups; assessment of the needs and requirements from management and users of those items. In general, this stage contains review of eight dimensions of the products/services: volume, prices, demand, suppliers’ competitiveness, evaluation criteria, restrictions, specifications, and processes. The extent of the review is mainly based on the available information. 13. The second step refers to the market analysis and is in essence a review of the supplying market. This stage is also strictly subject to the information that might be available but should include factors such as: local supply market structure, level of competition, laws and regulations, government purchase power, among other factors. 14. After having assessed the items, their related markets and operational context of the procurement of each item, the next step in the strategic sourcing model is the development of a procurement strategy that can generate savings. The parameters that guide this analysis 68 include the scale gains, price negotiation, expansion of the supplier base or optimization of the relationship with the current one and changes in the implementation of procurement procedures (Figure 2-4). Figure 2-4 Parameters to define a strategy for savings Source: World Bank. 15. The result of the previous analysis leads to the identification of different strategies for each of the families of items based on their specific characteristics. These strategies might be oriented to price reduction, which is generally a more conservative approach or the improvement of efficiency which is a more aggressive approach and requires deeper measures. This analysis suggests measures in addition to cost reduction for each of the three groups in order to find incremental measures that would yield additional savings (Figure 2-5). Figure 2-5 Definition of a strategy for savings Source: World Bank. 69 16. After the strategies have been assessed and defined, the implementation of the selected strategy will be developed. This step requires the confirmation of the key aspects needed to implement the strategies, such as commitment from key agencies, legal support, definition of roles and responsibilities among relevant government functions. Relevant information include reports on data, redefined bidding documents and the design of new processes. 17. This stage culminates in the agreement of the Government to implement the strategy. In general, the strategies shall be designed so that no law or decree be required for their undertaking. In this particular case, the incremental changes proposed for each of the items are a series of separate, although coherent, actions to be launched and implemented on a short term basis without making changes on the procurement legal framework. 18. Finally, the last milestone corresponds to the implementation of the strategies. It supports the Government during contract execution and includes evaluation of performance of suppliers through performance indicators, monitoring of savings and results, collection of feedback from end users and training and capacity building to government officials. The goal of the capacity building initiatives will be to create permanent conditions to implement the successive approaches as per the methodology described above. Fiscal savings from more efficient procurement strategies 19. The analysis undertaken by this review shows that more efficient procurement strategies yield fiscal savings of up to 0.4 percent of GDP. Achieving these savings requires a comprehensive reform of the procurement system for all of the more than 1,700 categories of goods, works, and services purchased by the general government of Peru. Two different scenarios were designed to estimate the potential savings that would result from the implementation of strategic changes in procurement. The “conservative” scenario assumes the implementation of innovative and modern procurement strategies that tries to leverage the purchasing power of the government in the market to attain lower prices and, in some cases, increase and standardize quality. Savings for this scenario are calculated based on prices currently paid by the government. The “aggressive” second scenario simulates that, in addition, assumes that the government adopts optimization initiatives to curb the current level of consumption and/or replace expensive with more cost effective inputs. The fiscal savings opportunities amount to between 1.1 billion Soles under the conservative scenario and 2.2 billion Soles under the more aggressive procurement strategy scenario. The detailed fiscal savings for each product category that the general government procures are listed in Table 2-1. The degree of complexity of the implementation of the reforms necessary to materialize the fiscal savings through more efficient procurement strategies varies among the different product categories that are procured; the degree of complexity is determined through the supply positioning tool described earlier (see Figure 2-2). 70 Table 2-1 More efficient procurement strategies yield fiscal savings of up to 0.4% of GDP Implementation Category Total Amount Potential Savings Complexity In Soles Conservative Aggressive 10-High; 1-Low CONSTRUCTION 7,793M 3% 6% 9 ADMINISTRATIVE CONTRACT FOR SERVICES 3,375M 6% 15% 6 MISC. SERVICES 2,067M 3% 8% 3 ROAD AND BRIDGES MAINTENANCE 1,353M 3% 7% 8 COMBUSTIBLES AND FUELS 549M 5% 10% 3 FOOD AND BEVERAGE FOR HUMAN CONSUMPTION 517M 5% 12% 3 COMPUTATIONAL AND PERIPHERAL EQUIPMENT 451M 31% 46% 3 SECURITY AND SURVEILLANCE SERVICES 406M 10% 16% 3 MACHINERY, EQUIPMENT AND FURNITURE IN OTHER 389M 11% 22% 4 INSTALLATIONS CONSULTING 340M 9% 15% 8 VEHICLES 538M 6% 12% 4 MAINTENANCE OF BUILDINGS, OFFICES AND 339M 3% 5% 7 STRUCTURES MEDICAL, ODONTOLOGICAL AND LABORATORY 337M 7% 14% 7 MATERIAL RENTING OF BUILDING AND OTHER STRUCTURES 299M 8% 12% 5 MEDICINES 282M 7% 15% 7 VEHICLES MAINTENANCE 274M 2% 9% 4 TELECOMMUNICATIONS EQUIPMENT 257M 12% 20% 4 CLEANING AND HYGIENE SERVICES 220M 10% 19% 3 TRAINING AND IMPROVEMENT SERVICES 217M 5% 8% 4 OFFICE SUPPLIES 273M 15% 25% 3 CEMENT 210M 3% 6% 4 PRINTED PUBLICATIONS 355M 4% 4% 4 OTHERS 5,974M - - OTHER SERVICES 1,341M - - TRAVEL, TRAVEL EXPENSES, PER DIEM 965M - - FLIGHTS 363M - - SECURITY AND PROTECTION EQUIPMENT 324M - - ADVERTISING SERVICES 210M - - INSURANCE 291M - - Total S/ 30,310M 1.1 M 2.2 M Source: MEF and WB estimates. 20. The supply positioning analysis further helped to identify ten families of goods and services that can easily be impacted by procurement strategies (Figure 2-6). These groups have been selected based on the following two criteria: a) they account for a significant share of the value of total expenditures and b) a procurement strategy over these items can potential generate short term savings without reform or long and complex changes of the current system. 21. The analysis undertaken by this review shows that straightforward changes in procurement strategies for these nine standardized goods alone would yield up to 257 million Soles in fiscal savings. The necessary changes in the procurement procedures can be implemented without any changes in current procurement laws or systems. For these nine items, savings opportunities in procurement would be around 139 - 257 million Soles each year (Figure 2-7). 71 Figure 2-6 Product families of items identified for strategic sourcing Source: World Bank analysis based on SEACE data in 2015. Figure 2-7 Potential savings on strategic items implementing procurement strategies (in million soles) Source: World Bank analysis based on SEACE data in 2015. Fiscal savings from strategic sourcing for vehicles, fuel, and cement 22. This section provides a detailed strategic sourcing analysis for three main families of products: vehicles, cement and fuel. As described, these items were identified based on the level of procurement and characteristics of the products that make them suitable to procurement strategies that can yield monetary savings in the short term. The assessment studies the main characteristics of these families of products, main buyer locations and quantities. Each of the sections finalizes with specific proposals to improve efficiency in the procurement of these items. 23. The potential fiscal savings from improving the public procurement for only three selected product families alone amount to between 57 and 115 million Soles. The conservative scenario, which assumes the implementation of low cost, short term measures, shows potential savings between 3 and 6 percent of the procurement expenditures for these three items (Figure 2- 8). Savings are largest for vehicles and fuel. The implementation of more comprehensive 72 procurement changes, as estimated by the aggressive scenario, would yield potential savings of at least ten percent for these two product families. Figure 2-8 Potential savings of applying procurement strategies to three products amount to between 57 and 115 million Soles each year Source: World Bank analysis based on SEACE data in 2015. Procurement strategy for vehicles 24. The procurement of passenger vehicles is largely carried out by the Central Government and Lima. Based on the analysis of data from 2015, eighty-six percent of the amount for this item has been procured in Lima and two main type of vehicles are purchased: sedans and SUVs. These characteristics facilitates the implementation of a procurement strategy in the short term and increases potential savings. 25. Despite having one main destination for the vehicles, substantial fragmentation on budgeting and procurement happens. Six hundred thirty-eight cars were purchased in 2015 but except for one large procurement process, all the rest have been acquired in very small batches, with seventy-five percent of the processes buying six cars or less at a time. 26. The fragmentation was particularly pronounced for Sedan cars. Except for one larger procurement, more than two hundred cars have been purchased in a scattered manner and in small volumes that resulted in very different specifications and prices. This fragmentation also resulted in small processes which resulted in less competitive procurement methods. In fact, the highest number of processes have been carried out through ADP (direct public awarding) and ADS (selective direct awarding) which are non-competitive methods (Figure 2-9). These results show a low level of competition; considering that 20 out of 28 processes had only one bid. 73 Figure 2-9 Procurement processes for sedan cars by method (number of processes and amount in soles) Source: World Bank based on SEACE data in 2015. 27. This fragmentation as well as low competition or even non-competitive processes resulted in a large dispersion of prices paid of almost six hundred percent for similar products (Figure 2-10). Some cases were identified where the same supplier charged different prices for a similar product depending on the purchasing entity. The dispersion in prices might to some extent be due to the diversity of technical specifications for the same item, as each entity prepared their own requirements despite the fact that they were buying the same goods. 28. The criteria applied for the selection of the lowest bid regarding the procurement of new sedan cars in the different entities is based on the acquisition price. However, for this kind of product, the initial acquisition price may represent only sixty-seven percent of the total life-cycle cost in the best scenario (Figure 2-11). This implies that in order to improve efficiency in the total life cost of vehicles, it would be important to take into account the whole cost of ownership of a vehicle. This approach would consider fuel consumption and maintenance costs as part of the evaluation of bids and award criteria. This evaluation method would analyze the price per mile of each car that is being offered, considering specially that the cost varies depending on the model and brand. Figure 2-10 Unit price dispersion in sedan cars (in Soles) Source: World Bank based on SEACE data in 2015. 74 Figure 2-11 Estimated vehicle life-cost (in %) Source: World Bank based on Automotive Fleet. 29. The procurement of SUVs shares shows similar characteristics to the sedan market. There is a substantial variation of the technical specifications between entities that had an impact on prices producing a large dispersion (Figure 2-12 and 2-13). Regarding quantities, the three hundred forty-three wagons procured in 2015, were purchased in small quantities except for one large procurement process. The evaluation criteria do not include life cost maintenance even when the purchase price of this kind of product can be as low as sixty-four percent of the total cost of operation. 30. The competition was also very low for this kind of item where ninety-seven percent of the procurement processes had only one or two bids. Most of the procurement processes of SUVs have been non-competitive and the type of technical specifications in these cases has been more sophisticated. This ended up getting more expensive prices than the ones obtained using open procurement methods, which require publication of the specifications. As figure 2-13 shows, this dispersion has even been identified for the same type of SUVs. Figure 2-12 Unit price dispersion for SUVs (in Soles) Source: World Bank based on SEACE data in 2015. 75 Figure 2-13 Unit price dispersion for same type of SUVs – 4x4 (in Soles) Source: World Bank based on SEACE data in 2015. 31. Considering the main results of the procurement processes carried out for vehicles, and specifically for sedans and SUVS, the following alternatives are proposed to address the issues that were identified:  Planning: Fragmentation shows a lack of planning for the procurement of vehicles. In this sense, it is proposed to carry out an analysis of the annual demand of each entity and consolidate the most of it on larger contracts,  Technical Specifications: Standardized technical specifications in categories based on their use avoiding that each entity defines their own requirements based on other criteria (such as design),  Evaluation criteria: Include evaluation criteria that considers life-cycle cost of the vehicles in order to award bids using the cost per kilometer or mile,  Procurement methods: Based on the planning exercise, larger procurement processes would be carried out avoiding the use of non-competitive procurement methods. This type of methods tends to end up getting higher prices.  Publishing: Increase competition through the delivery of invitations to different type of suppliers,  Other alternatives: For the long-term planning, it is advisable to analyze the cost of leasing against the procurement and maintenance costs. 32. The recommendations are oriented to the implementation of incremental changes on the current procedures demanding small adjustments. The application of these alternatives would have a strong impact on the enhancement of efficiency and mainly in the reduction of the current dispersion of unit prices and transaction costs. As described previously, the implementation of these measures would result in different savings scenarios based on the type of alternatives applied considering more aggressive or conservative changes. Procurement strategy for fuel 33. Purchase of fuel by the government of Peru is highly fragmented, even more than vehicles. The four largest buyer entities of fuel in 2015 accounted for sixty percent of the government demand for fuel. However, the data shows that out of forty-six million gallons of diesel that have been purchased in 2015, nearly 3,000 processes were carried out. From this total 76 amount, seventy-five percent of the processes have been carried out for quantities below 8 thousand gallons. 34. As analyzed for the other families of products, the fragmentation of the procurement of fuel caused a high number of processes carried out through less competitive procurement methods that resulted in low competition and had a strong impact on prices . The highest values have been contracted through AMCs, which accounted for forty-one percent of the volume of procurement processes (Figure 2-14). This has resulted in very low levels of competition; where seventy-six percent of the processes had less than three bids. Figure 2-14 Fuel unit price by procurement method (in Soles) Source: World Bank based on SEACE data in 2015 35. However, Loreto and Arequipa, have made high volume purchases that have shown lower unit prices than the average, confirming that strategies to leverage demand yield lower prices for the government. In the case of Arequipa, fifty-one percent of the total quantity procured got a unit price of S/8.68. In Loreto, the unit price was S/6.90 for seventy-seven percent of the purchased volume (Figure 2-15). This approach exemplifies how centralized strategies can result in substantial savings without major changes in the current system. Figure 2-15 Quantity of fuel procured in 2015 and average unit prices by department (in soles) Source: World Bank based on SEACE data in 2015. 77 36. Regarding supply, the data shows that the main supplier of fuel has been Petro Peru offering the lowest prices and twenty-three percent of the procured total amount . Due to the fragmentation of demand, the rest of the suppliers have offered smaller quantities and higher prices. 37. Considering the main results of the procurement processes carried out for fuel in different departments in Peru, the following suggestions are proposed for further consideration:  Planning: Based on the fractioning that was identified, it is recommended to carry out an analysis of the annual demand of each entity and consolidate the most of it on a single contract at the regional level. The experiences of Loreto and Arequipa should be replicated as a best practice within Peru as they how clearly the impact of larger procurement processes on prices and efficiency,  Evaluation criteria: As it was mentioned in the previous section, it should be encouraged the inclusion of fuel consumption as additional criteria for the evaluation of bids in the procurement of vehicles or other associated goods. This would also have a strong impact on the consumption of fuel in the near future,  Procurement methods: The improvement of planning would demand the execution of larger procurement processes. This is recommended in order to avoid the use of non-competitive methods or procedures intended for small purchases. Small procurement processes should be limited to emergency situations,  Other alternatives: Regarding the consumption of fuel, there is an alternative to explore the use of a telematics tool to manage the fleet of vehicles and monitor life cost of these goods. This tool supports the improvement of efficiency through monitoring the government’s fleet of vehicles and their associated consumption. It is a multidisciplinary area encompassing telecommunications, vehicular technologies and computer science that includes technologies for sending, receiving and storing vehicle information and data. Some studies have shown that the application of these kind of technologies have an impact on fuel consumption, accidents and speed driving. 38. The implementation of strategies oriented to the reduction of unit costs or improvement of efficiency would have a strong impact on the current expenditure in fuel consumption. A conservative scenario shows a reduction of five percent in the current amount and an aggressive scenario of ten percent reduction. Procurement strategy for cement 39. Procurement of cement is also highly fragmented and supplied locally . Specifically, during 2015, two thousand ninety-one procurement processes were carried out for small quantities. As a result of the large volume of small processes, the variation in prices paid for cement is tremendous and the highest prices have duplicated the average paid by the government. 40. As a result of the fragmentation in the procurement of cement, a total of eight hundred thirty-two different suppliers have been awarded supply contracts . Besides this, the prices paid by each of the departments have varied substantially even in neighboring areas. Based 78 on the data that was analyzed, it was identified that even the same supplier has charged different prices within the same department. 41. There have been processes to obtain cement in Peru during 2015 that show the benefits of larger procurements. The largest purchase was carried out in Puno and the price was S/20.60 soles by unit. Also in two other large purchases that were analyzed, the unit cost obtained was below the target price for the calculated savings (Figure 2-16). Figure 2-16 Unit costs for the procurement of cement in 2015 (soles) Source: World Bank based on SEACE data in 2015. 42. Considering the lack of coordination and its impact on prices, it was estimated with some articulation between regions almost eighty-four percent of the total value could be carried out as a regional procurement process. A higher level of coordination with the central government or even between neighboring departments, would have an impact on the prices obtained (Map 2-1 and Figure 2-17). Figure 2-18 shows the geographic concentration of the procurement of cement that could potentially be carried out in a single process. Regional procurement processes would have an impact not only on unit prices but also on reducing the transactional costs of carrying out a high number of processes. 79 Map 2-1 Average unit price by department (in Soles) Source: World Bank based on SEACE data in 2015. Figure 2-17 Average unit price and percentage bought Source: World Bank based on SEACE data in 2015. Figure 2-18 Distribution of the procurement of cement by region (in Soles) Source: World Bank based on SEACE data in 2015 43. Non-competitive methods and limited bids, have resulted in higher prices. Similar to the other items, almost 50 percent of the procurement processes of cement have been through non- competitive methods, which have resulted in higher prices. Regarding the level of participation in procurement processes, sixty-four percent of the procurement of cement has had less than three bids. This is directly related with the type of procurement method that is applied. There is a direct relationship between the number of bids and the unit price, showing higher prices on less participation (Figure 2-19). In average, the lowest unit price has been obtained from the electronic 80 auction. If this price was obtained for the whole number of processes, the government would have saved more than 14 million Soles. Figure 2-19 Unit price by procurement method and number of bids (in Soles) Source: World Bank based on SEACE data in 2015. 44. Considering the main results of the procurement processes carried out for cement in different departments in Peru, the following alternatives are proposed for further consideration:  Planning: Based on the fractioning that was identified for the different departments and the impact on unit prices, it is recommended to carry out an analysis of the annual demand of each department and consolidate the most of it on regional purchases. The analysis has shown there is potential to merge certain volumes which would significantly reduce unit prices.  Procurement methods: As it was proposed for the other items, it is advisable to avoid the use of less competitive methods and encourage the application of higher competition to gain better unit prices,  Other alternatives: For the long term on the procurement of cement, it is advisable to carry out a detailed analysis of the works contracted using this cement to get information on performance, quality, quantities that would serve as an input for future planning. 45. The recommendations set out to improve efficiency in the procurement of cement are mainly oriented to the reduction of unit prices. The analysis of the current data shows a conservative scenario of a reduction of three percent in the current expenditure by reducing the unit cost. However, the aggressive scenario encourages the research on how to enhance the use of cement with efficiency measures. Long-term fiscal savings 46. After identifying the main procurement categories of the government of Peru, the previous sections have carried out a strategic sourcing analysis for three of the main families of products procured. Based on the analysis of public procurement data from 2015, the following section identifies additional areas that strengthen Peru’s long term procurement strategy and proposes best practices to address them. 81 47. The government of Peru carries out procurement processes for a large and diverse variety of goods, works and services with different levels of complexity. In 2015, there were 17,583 families of items that were procured. Only a few of them absorb the majority of the money spent.18 Designing a specific strategy for those families of items for which the government procures high amounts, which should be different from the one that represent lower amounts, is recommendable. Improving the length and timing of procurement procedures 48. In 2015, almost forty-two percent of the total amount purchased by the government of Peru was contracted during the last quarter. In central and regional governments, about a quarter of the amount has been contracted in the following year. The level of procurement during the first half of the year has been low: thirteen percent on average for all levels of government and 11 percent for the central government (Figure 2-10 and 2-21). Also, only twenty-seven percent of the total amount that was awarded had been actually published during the first semester. The length of procurement procedures also varied significantly (Figure 2-22). Figure 2-20 Procurement implementation timing during 2015 Publishing Contract signature Source: World Bank based on SEACE data in 2015. Figure 2-21 Procurement timeframe during the year by level of government Source: World Bank based on SEACE data in 2015 18 Three of these families of products were analyzed in the previous sections. 82 Figure 2-22 Length of procurement process by method (in days) Source: World Bank based on SEACE data in 2015. Avoiding fragmentation of procurement management 49. As it was identified for the three families of items analyzed, the public procurement system in general shows high levels of decentralization. The data indicates that there are more than 2,800 purchasing entities. Many of them have procured very small amounts in total, in fact, half of them bought less than 1.9 million soles in the entire year (Figure 2-23). This fact, along with the lack of coordination between entities, does not allow the possibility to plan and carry out larger procurement processes. 50. In line with this, a large volume of processes was carried out for reduced number of families of goods that are periodically bought by the government. These data allow to infer a strong need for an adequate strategy for the procurement of these items carried out by multiple entities. This change would reduce administrative costs and bureaucratic steps, which would provide the possibility of orienting human resources to work in other tasks different from the procurement sector, which are more necessary to the services provided by the State (such as health, education, security, etc.). 83 Figure 2-23 Number of entities that bought these families and percentage of the total procurement (2015) Source: World Bank based on SEACE data in 2015. Market concentration and diversity of unit prices 51. Some of the items procured by the government of Peru during the year 2015 were provided by a limited number of suppliers and in some cases by a single one. Taking into consideration the high decentralization of the procurement management, the result is a large number of processes that end up being awarded to the same supplier or a restricted group. Cases have also been identified in which the same supplier has charged different prices to various purchasing entities. This has two kinds of costs for public resources: higher unit costs paid by the entities and unnecessary administrative costs to carry out such a large number of processes. 52. Several items have few suppliers and need specific procurement strategies. Items with a low level of supplier diversification (Figure 2-14 in red bubbles). These items require a specific procurement strategy because of the characteristics of the market in which they operate. In particular, it is the pharmaceutical sector that presents monopolistic or oligopolistic markets mainly due to the existence of patents. Figure 2-24 Number of suppliers and diversification index (2015) Source: World Bank based on SEACE data in 2015 53. The following example on levotiroxina illustrates the analysis described on market concentration. Levotiroxina is a pharmaceutical that was procured by ten entities during 2015 for a total amount of 7,664,733 soles. Thirty-eight procurement processes were carried out for this item and thirty-six ended up being awarded to the same supplier. Unit costs declined when higher 84 quantities where procured (Figure 2-25). As a conclusion for this specific item, if this characteristic was identified and addressed properly with a procurement strategy, the purchase would have been carried out in less procedures and probably getting lower prices. Figure 2-25 Example on levotiroxina Market concentration for a specific item and impact on prices Amount (Soles) Buyer entities Number of processes Total 7,664,733 10 38 Merck Peruana S.A. 7,607,508 8 36 Representaciones Deco S.A.C. 40,500 1 1 Pharmintegra S.A.C. 16,725 1 1 Source: World Bank based on SEACE data in 2015. Towards more competitive procurement methods 54. Procurement processes carried out using the AMC and ADS method represent eighteen percent of the contracted amount but seventy-eight percent of the volume of processes. These numbers suggest a highly transactional cost to execute a smaller part of the budget for public procurement through the use of procurement methods for minor amounts. Only AMCs account for fifty-two percent of the total number of executed processes (Figure 2-26). Figure 2-26 Number and amount of procurement processes based on method (in soles) Source: World Bank based on SEACE data in 2015. 55. The number of procurement processes reduces drastically after the amount of 35,000 soles is reached (Figure 2-27). This fact might suggest that there is fractioning in the contracts in order to use the threshold for AMC. 85 Figure 2-27 Number of procurement processes based on the amount Source: World Bank based on SEACE data in 2015. 56. The electronic auction modality has resulted in better participation receiving the higher number of bids, followed by public tenders and ADSs (Figure 2-28). The data analyzed for 2015 also shows that almost eighty percent of the procurement processes had less than three bids. Figure 2-28 Average number of bids by procurement method Source: World Bank based on SEACE data in 2015. 57. Ninety-six percent of the procurement processes performed during the year 2015 failed in a first attempt, which seems a systemic problem of Peru’s procurement system. This result is significant at all levels of government (Figure 2-29). This figure is highly relevant and presents opportunities for developing a clear procurement strategy. A process that is not successful has major consequences mainly on the lack of availability of the good, work or service to be procured, but also with a high administrative cost that uses additional resources that could be applied to other areas of government. On top of this, the failure of a process allows the second attempt to be made according to the rules of acquisition of processes with smaller amount. This could explain the high volume of processes developed by this method and the price consequences that were analyzed previously. In this context, and as evaluated in the previous sections with respect to the number of bids presented, it could be inferred that the failure of the processes may be caused in part by the lack of interested bidders but also by the other areas of improvement identified during the analysis. 86 Figure 2-29 Processes failed by level of government Source: World Bank based on SEACE data in 2015. Conclusions 58. The main finding of this exercise is that there is a great deal of opportunity to improve efficiency and performance of the procurement system in Peru. In fact, the government procures more than 1,700 different categories of goods, works and services and yet it relies on the same set of procedures and tools to procure all of them. Consequently, there is ample room to develop procurement strategies that are fit for purpose and specific to improve results on different markets. There is a promising and untapped opportunity to materially improve results that the government is currently getting on procurement. Suggestions to improve these results are discussed below and presented into three groups: first, systemic issues and areas where procurement underperforms in Peru; secondly, short-term ideas to tap on low-hanging fruits and to achieve monetary savings through better procurement strategies for the three categories discussed in this report – these are the measures proposed to achieve the conservative savings scenario presented earlier; finally, a set of medium-term ideas to craft procurement policy that will produce material savings for the three categories discussed in this report. This corresponds to the aggressive scenario presented earlier. 59. Analysis of procurement data revealed three important areas where the Peruvian system is currently underperforming: competition for government contracts, success of bidding processes - measured by the number of processes that result in a contract - and strategic approach to procurement planning and implementation. According to the data, poor or total lack of competition for government contracts is a systemic issue that is currently undermining results for the procurement system. Poor competition is manifested in two forms: (1) a very low turnout of bidders – almost 90 percent of all successful bidding processes had three bidders or less and 60 percent had only one bidder; and (2) the use of less competitive methods or direct contracting – 31 percent of the awarded value and almost 90 percent of volume procured did not follow competitive procedures. 60. A concerning 96 percent of all bidding processes failed to result in a contract at the first attempt. Failed bidding processes have practical consequences for the results obtained by the government. For instance, a failed bidding process will have to undergo rebidding, which adds a 87 significant amount of time until the goods, works and services are finally available to deliver services to communities. In addition to delays, a regulatory provision in Peru allows a rebidding to follow a less competitive or direct award method, which further undermines the results for the government. Finally, the administrative costs of processing more than 106 thousand failed processes were certainly substantial. 61. The Peruvian procurement system currently lacks a strategic approach to procurement, in the sense that processes and procedures are applied the same way to very different markets. The Government of Peru procures a very wide range of different goods, works and services of varying complexity and risk to the government service delivery chain. Yet, the approach to procurement does not consider these elements of complexity and risk and neither the level of demand and the market structure. This poses a relevant opportunity to craft policies and strategies that are fit for purpose and thereby increase the chances for better results in terms of price, administrative efficiency, timeliness, quality and transparency. The absence of procurement strategies can be seen in the data as there is excessive fragmentation of government demand for certain types of goods, works and services where the government is a major player in the market and could leverage this position to attain better results. In this report, we discussed three categories of products where the volume purchased by the government merits a more strategic approach to the market. We also showed isolated examples of better coordination of procurement even at local level, resulting in lower prices. Thus, the potential for more centralized strategizing is certain to result in better prices if properly done. And better strategies are likely to reduce the administrative burden and cost of discharging the procurement function at government agencies. This is a major hidden cost for the government – as a matter of fact, a total of 1,000 different government agencies had to maintain a structure to carry out procurement in 2015 when they only procured less than 1 million Soles. Better procurement strategies could minimize the administrative footprint of procurement processes and improve resource allocation for the government. 62. To address these systemic issues, a careful reform process that combines regulatory changes, policy development, capacity building, enhanced oversight and computer infrastructure will be necessary. This means that material improvements for the overall system is a medium term endeavor that will need serious and committed leadership. Short-term but impactful policy options 63. But there is a wide range of options to make significant improvements to procurement in the short-term. These can be catalytic for a broader reform to the overall system. In this report, it was discussed three categories of products – fuel, cement and vehicles – in which major improvements can be easily achieved without the need for difficult and time-consuming changes to the current system. 64. The conservative savings scenarios presented earlier in this report for fuel, vehicles and cement do not demand any major regulatory changes: just a different way of implementing procurement within the current regulatory boundaries. To achieve the conservative savings scenarios, a combination of measures based on strategies specific to the market will be all that is needed. More specifically, better procurement planning that breaks the silos of individual agencies and look at aggregate demand by the government as a buyer is critical. 88 Moreover, the use of competitive procurement processes is another requirement thereby limiting the current large volume of uncompetitive awards. To dovetail with more competitive procurement procedures, it will be necessary to reach out to suppliers and businesses of these specific markets to inform the new approaches. It is also possible to streamline requirements to prepare a bid in Peru, and this tweaking to implementation can make an important contribution to competition and reduce the cost of doing business with the government. 65. Consolidation of demand at appropriate level will be critical to achieve the estimated conservative savings. As discussed earlier in the report, procurement can be implemented at central, regional or local level and for cement, for instance, there seems to exist an opportunity for cross-regional collaboration that has the potential to benefit multiple government agencies with lower prices and better quality. 66. More meaningful evaluation criteria should be part of the procurement strategy for cement, fuel and vehicles. And this does not require any change in current regulations. As discussed, new vehicles are currently procured based on its purchase price tag but the cost of operating these vehicles encompasses additional elements, such as maintenance and fuel consumption. This report recommended adopting price per mile or price per kilometer as a more meaningful way of evaluating bids and awarding contracts for the government. This approach is currently used by many businesses that show a high-performing fleet management system. Medium-term policy options 67. In addition to the above short-term measures, this report proposes a set of medium- term policy options to unlock more significant savings in the procurement of fuel, vehicles and cement; we refer to these options as the aggressive savings scenario. In addition to better procurement strategies and approaches, the Government of Peru could also review the current levels of consumption of fuel, vehicles and cement. Besides, in the case of passenger vehicles, the diversity of specifications results in very different vehicles being purchased for the same purpose. Naturally, prices vary significantly as well. 68. The development of procurement policies regarding consumption and specifications will unlock major savings for the Government of Peru. On the demand side, policies that tighten requirements for buying an additional vehicle would have an important impact. In 2015, almost four thousand passenger vehicles in addition to almost two thousand motorcycles were introduced to the government’s fleet. The addition of a vehicle to the government’s fleet brings with it maintenance, repairs and fuel needs. The Peruvian Government spent more than 240 million soles in vehicle maintenance and repairs and almost 550 million soles in fuel in year 2015. Thus, there should be appropriate policy requirements and a justified business need before an agency can even start the procurement of a vehicle. 69. And the procurement of a new vehicle should try and leverage the demand of the government as a buyer. In this context, agencies should be required to follow centralized approaches unless they can justify and prove that they were able to get a better deal on their own. Regulations and policies making central procurement arrangements mandatory would curb 89 unnecessary waste and prevent excessive fragmentation of demand, which data shows results in less competitive procurement, higher prices and over-specification of products. 70. On the same vein, the government would certainly benefit from policies and regulation for fleet management. With the addition of more than 6,000 passenger vehicles every year to the fleet, it seems important to develop policies and procedures for (1) buying a new vehicle, (2) maintaining and (3) repairing it. Data shows that government agencies purchase very different vehicles for the same purpose – even within the same sector – revealing a lot of discretion for buyer’s choice. This inevitably results in a wide range of prices paid for vehicles that will serve the same purpose and at the same time it increases significantly the costs for maintenance and repairs as agencies will have to deal with multiple suppliers. This report also suggested the introduction of telematics to support a fleet management policy and to build a database and information about vehicles owned by the government. This technology can also leverage control over the fleet that has proven to reduce fuel consumption and to improve driver behavior, which results in less accidents and speeding. The amount of money spent by the Peruvian Government in vehicle warrants a more detailed and centralized policy for fleet management. 90 Part II – SECTOR ANALYSES Introduction 1. Peru has among the highest investment spending in water and sanitation (WSS) in the region but access to piped WSS services remains below that of its regional and structural peers, especially in rural areas. Adequate drinking water and sanitation are critical for human health. Yet, about half of Peru’s rural population does not have access to improved water and sanitation services. This is significantly below access in structural and regional peers. Lack of access is not the result of too little investment. But investment has only translated into limited improvements in access: Peru spent an average 0.8 percent of GDP between 2010 and 2014 on WSS investment but access to piped water increased only from 77 to 81 percent and access to sanitation from 56 to 66 percent between 2009 and 2014. 2. Improving public investment management and the performance of water public utilities could lead to fiscal savings of at least 0.3 percent of GDP and help to accelerate access to WSS services. The WSS sector suffers from deep-rooted inefficiencies that arise from lack of incentives to prioritize investment projects efficiently and in line with sector priorities; and properly fund the operation and maintenance of existing WSS infrastructure. Implementing a multi-annual public investment framework with clear priorities, bundling investment projects, improving O&M spending and services provision while raising water tariffs, would not only yield significant savings, but could also put the Sustainable Development Goals of universal access to improved WSS services within reach. 3. This chapter assesses WSS spending patterns and identifies inefficiencies in the WSS sector with the objective to identify policy options to improve the sector’s performance. The second part describes the spending patterns and compares them with the spending patterns of regional and structural peer countries. It also describes the outcomes of sector investments, presenting key indicators measuring the access to and quality of water and sanitation services. The third part highlights areas where inefficiencies in the sector can be identified that are related to the lack of incentives for utility companies to improve their operational and financial performance and manage public investments.19 The final section provides an overview a key policy options to address the large inefficiencies in the sector. Spending and Outcomes 4. The spending analysis focuses on capital spending since reliable data on total WSS spending is not available. The WSS sector in Peru involves many institutions (see Box 3-1). In many small cities and communities attributing staff time to WSS spending is complicated as staff 19 The analysis of this chapter has been constrained by lack of data. First, data on current spending in the sector is likely incomplete, requiring the analysis to focus on capital spending. Second, data on public spending by subsector (water supply, sanitation, wastewater collection, wastewater treatment, and supporting activities) and type of activity (expansion, rehabilitation, replacement, or technical assistance) is not available. Third, data on WSS companies, except for water utilities in urban areas (EPS), is not available. 91 is likely to be engaged in other activities than providing only WSS services. Hence, the classification of current expenditures mostly includes goods and services and, to a much lesser extent, staff costs and allowances. In fact, recurrent expenditures from SIAF explain on average only 3.2 percent of total WSS expenditures between 2009 and 2015, while WSS investment averaged 97 percent. Although the WSS sector tends to be capital intensive, this composition is highly unusual—normally current expenditures are in the range of 10–20 percent of total expenditures. Capital spending includes investments for expansion, improvement (or extension) of existing infrastructure, and rehabilitation. In the National Investment Plan for 2014–2021, the 2014 budget included a total of S/8.3 billion in investment, of which S/1.9 billion is for ampliacion, S/1.3 billion for mejoramiento, and the remainder of S/5.1 billion for both ampliacion and mejoramiento. Box 3-1 Water Sector Institutions in Peru The WSS sector in Peru involves many institutions. There is 1 sectoral ministry, 26 regional governments, and 1,965 municipal governments. In addition, the sector has a regulatory agency (SUNASS), 50 water and wastewater companies (EPS), service providers in small cities (between 2 and 30,000 inhabitants), and Boards of Directors of Sanitation Services (Juntas de Administracion de Servicios de Saneamiento, JASS) in rural areas (see Annex 1 for a detailed overview of the sector). The Ministry of Housing, Construction and Sanitation (MVCS) establishes policies and programs at the national level and finances investment projects. It also decides on the capital transfers to regional and local governments and EPS. In some instances, it directly executes investment projects, especially in rural areas. The Ministry does not determine current spending for administration, operation, and maintenance. The Ministry of Economy and Finance (MEF) – MEF’s Directorates of Public Budget (DPP) and Public Debt (DCP) determine to some extent the financing policies of water utilities. They do so by setting limits on the water utilities’ indebtedness and by guaranteeing their debt to local and international lenders. In practice, MEF has been paying most local WS&S utilities’ debt obligations to bilateral organizations; it has also paid or pardoned water utilities’ tax liabilities. Attached to the MVCS is the Technical Organization of the Administration of the Services of Sanitation (OTASS), a public agency that among other national competencies is in charge of administrating and managing the EPS, which are key providers of WSS services in urban areas. The EPS consist of 51 utilities organized under private company law and owned by locally elected municipal governments, except in the case of Lima, where the central government retained ownership and created SEDAPAL. The economic regulator is the Super Intendencia Nacional de Servicios de Saneamiento (SUNASS) . Set up in the early 1990s as an independent economic and quality of service regulator, SUNASS mandate is to set tariffs consistent with consumers’ willingness to pay and with the financial viability of the water utility companies. In principle, tariffs are linked to the water utilities’ investment programs and to their progress in implementing these programs. 5. WSS capital spending increased rapidly in recent years, driven by higher investment spending in rural areas and increased capital transfers to urban WSS providers. Capital spending nearly doubled from S/2,222 million in 2009 to S/4,379 million in 2014 (Figure 1a). Investment spending in rural areas nearly tripled during this period. Capital transfers, which consist mostly of spending for pre-investment studies, technical designs, and capacity programs, as well as project implementation costs, also increased significantly. They amounted to 6.6 percent of total capital spending in 2009 and increased to more than 16.7 percent of capital expenditure by 2015 92 (Figure 3-1.b). There has been a tendency to provide these transfers directly to public agencies, such as water and sewer companies (EPS), rather than regional and local governments. 20 Figure 3-1 WSS Capital Spending a. By Rural and Urban WSS b. By Spending Category 2,500 100% 90% 2,000 80% 1,500 70% 1,000 60% 50% 500 40% 0 30% 20% 2009 2010 2011 2012 2013 2014 2015 2009 2010 2011 2012 2013 2014 2015 10% Rural WSS Urban WSS 0% 2009 2010 2011 2012 2013 2014 2015 Infrastructure investments Purchase of other fixed assets Grants and Transferts Intangible investments Urban WSS Rural WSS Transfers Project preparation Other Non-financial assets expenditure Source: SIAF, author’s calculations. 6. The share of rural WSS capital spending has significantly increased in line with the National Sanitation Plan 2006-2015. Per capita rural WSS capital spending also surpassed urban WSS spending. Between 2009 and 2014, urban WSS capital spending climbed from Soles 83 per capita to Soles 100, while rural WSS capital spending increased from Soles 67 per capita to Soles 262. The increasing role of rural WSS in overall capital spending is consistent with the policies and targets set in the National Sanitation Plan 2006–2015, which aims to reduce the gaps in WSS coverage between rural and urban areas. 7. The average cost of WSS infrastructure is higher in rural areas. In most countries, the average unit cost of supply in rural areas tends to be lower than the cost of supply in urban areas. It is the opposite in Peru, which may be due to the lack of economies of scale due to significant project fragmentation (see below). For example, due to the larger share of small projects in rural areas and high project preparation costs due to SNIP requirements, the share of project administration costs was on average 6.4 percent of total capital spending in rural areas compared to 3.6 percent in urban areas. 8. WSS spending has become an important part of local government spending. By 2015, local governments spent more than 11 percent of their budget on WSS, which compares to only 1.4 and 1.2 percent for national and regional governments, respectively. WSS capital spending has increased for local and central governments, but declined for regional governments (Figure 3-2). 20 They are not defined as such in the budget and expenditure data in the SIAF. EPS may also receive funding directly through the regional and local governments. 93 Figure 3-2 WSS Spending by Level of Figure 3-3 WSS Spending by Source of Government Financing 5,000 4,000 3,000 2,000 1,000 0 2009 2010 2011 2012 2013 2014 2015 Local Government Regional Government National Government Source: SIAF, author’s calculations. 9. Funding sources for WSS infrastructure projects vary by level of government . For central governments, the largest source of funding are central government transfers ( recursos ordinarios). In fact, recursos ordinarios are the most important financing source for WSS capital spending (Figure 3-3). Regional governments also rely on central government transfers. Municipalities finance WSS services from a wide range of funding sources: canon y sobrecanon, regalias, renta de aduanas, y participaciones contributed almost 29 percent of WSS spending in 2015; funds from the Municipal Compensation Fund ( Fondo de Compensacion Municipal, FONCOMUN) accounted for 22 percent; municipal taxes also contributed to WSS funding. The high variability in the funding resources for oversight, feasibility analyses, and institutional support is a source of concern given the critical role in building an investment pipeline, providing technical quality control, and ensuring the quality of investment expenditures. 10. The recent decline in WSS spending, especially at the local government level, will slow down the progress towards universal coverage in water supply or sanitation. After years of increases, WSS budget allocations declined in 2015, especially for local governments. In 2015, WSS expenditure contracted by 26 percent to Soles 3.7 billion or 0.6 percent of GDP, which is linked to a decline in canon and other payments. The decline in canon payments, which started in 2012 may have been masked in recent years by delays in project execution. 21 11. Peru’s WSS investments were among the highest in the region and yet, Peru’s access to WSS services lags behind its structural and regional peers . In 2013, for example, Peru spent 0.7 percent of GDP on WSS investment, compared to 0.25 percent in Mexico and 0.6 percent in Colombia (Figure 3-4). Yet, access to piped water increased only from 77 to 81 percent and access to sanitation from 56 percent to 66 percent between 2009 and 2014. Relatively fast population growth in some provinces poses a challenge for expanding access to WSS services, while coverage has increased in provinces with a steeply declining population. 21 Contrary to government transfers, canon payments remain with the local governments. 94 Figure 3-4 WSS Sector Investment, 2013 0.80% % of National GDP 0.70% 0.60% 0.50% 0.40% 0.30% 0.20% 0.10% 0.00% Paraguay Uruguay Bolivia Mexico Nicaragua Colombia Chile Argentina Costa Rica Peru Panama El Salvador Ecuador Guatemala Brazil Source: Cepal, 2014. 12. Although WSS investments were higher than envisioned, they have not translated into significant improvements in access. The National Sanitation Plan 2006–2015 targeted an increase in water supply coverage of 83 percent in 2015 and improved sanitation coverage of 77 percent. It assumed that a total investment of US$4 billion (or Soles 11–12 billion) between 2006 and 2015 would be required to meet the sector objectives. Yet, between 2009 and 2015, total capital spending was nearly twice as high at almost Soles 21 billion. Nevertheless, the targeted outcomes of the National Sanitation Plan 2006-2015 have only been partially met. By 2015, improved water supply coverage stood at 88 percent but only 72 of the total population had access to improved sanitation. The improved sanitation coverage improved but fell short of the envisioned target, especially in rural areas. 13. Spatial disparities in access to adequate water and sanitation are still significant. Peru is behind its structural and aspirational peers in terms of access to water and sanitation, particularly in rural areas (Figure 3-5). While 88 percent of households nationally have access to water, this figure drops to 60 percent for households in rural areas. Likewise, sanitation coverage is 72 percent nationally, but only 22 percent in rural areas. 22 Open defecation is close to 30 percent among the rural population. Moreover, there is a large gap between urban, peri-urban, and rural areas in the quality of basic service provision. For example, for rural households, piped water coverage increased from 36 percent in 2004 to 65 percent in 2015, yet only 16 percent had access to water of drinking quality and fewer than 5 percent had access to water with some level of chlorine in 2015, well behind urban households. Even though urban areas show significantly better coverage levels, in fast-growing peri-urban areas WSS services tend to lag due to unplanned urban expansion and a lack of financial resources. As a result, households in these areas often pay much higher fees for access to these services. For instance, water can cost up to Soles 65 per 5 cubic meters in informal settlements, compared to Soles 12 for a similar amount of piped water in a residential area connected to the piped network.23 Continuity of water service also varies significantly. For instance, in Trujillo, Peru’s third most important city, average service is only 17 22 ANA (2015). 23 El Comercio (2015). 95 hours per day, and in smaller cities and rural areas it can be as low as only one to two hours per day.24 Figure 3-5 Access to water and sanitation is laggign behind peers countries Water Urban 100 80 60 40 20 Sanitation Rural 0 Water Rural Sanitation Urban Peru Structural peers Aspirational peers Source: World Development Indicators (WDI). Note: Structural Peers: Ecuador, Colombia, Mexico, Thailand, South Africa, Romania and Malaysia. Aspirational Peers: Poland, Korea Rep. and Chile. The data are for 2015. 14. Several key performance indicators of WSS Services providers have only improved in selected cases since 2005. Many WSS service providers still do not provide 24 hour services and metering and active connections fall significantly short of key targets. The operation margin of many providers has declined. Significant differences in access to improved WSS services persist across Peru’s regions. Piped water supply coverage, for example, is about 35 percent in Loreta but nearly 90 percent in Cusco and Arequipa. Inequality of access to improved WSS services has increased at the provincial level. 25 Table 3-1 Progress in Management Indicators as per National Sanitation Plan: EPS Only Indicator Actual 2010 target 2015 target Actual performance 2005 Performance EPS, 2014 Hours of Supply (hours/day) 17 23 23 16.2 Water Production (liter per capita per 291 230 200 235 day) Metering (%) 54 92 95 58 Account receivables (months) 5 2 2 Operating Margin (%) 28 42 50 10 Active Connections (%) 89 95 95 90 Source: SUNASS, National Sanitation Plan 2006–2015 24 SEDALIB S.A. (2015). 25 In 2009, 13 of the 25 provinces had water supply coverage rates above the national average; by 2014, only 10 provinces were above average. Provinces that are lagging behind the national average in piped water supply also do so in access to piped sewer services. Most households in rural areas depend on latrines, which have increased, partially funded by households. 96 Figure 3-6 The Marginal Cost of Expanding Access to Improved WSS Services a. Access to Improved Water Source b. Improved Sanitation Facilities 14100 2011 international dollars spent per additional connection with 2011 international dollars 12100 connection with access spent per additional 10100 10100 8100 access 5100 6100 4100 100 2100 80% 90% 100% Access 100 Bolivia Brazil Chile Colombia Costa Rica Ecuador 65% 75% 85% 95% El Salvador Guatemala Mexico Access Nicaragua Panama Peru Source: Staff calculations based on the CEPAL 2014 Source: Mexico Public Expenditure Review and World Bank World Development Indicators. Note: Chile’s 1999 WSS spending of 5.3 percent of GDP is excluded as an outlier. 15. The government needs to address these inefficiencies before expanding public expenditures. Improving the performance of the sector is particularly important since the cost of each additional connection tends to increase exponentially when approaching universal coverage (Figure 3-6). The last disconnected consumers tend to be located in especially remote or isolated areas, which can make the required additional connections very costly. Marginal costs of additional connections are also impacted by each country’s geographic and demographic characteristics. For example, Chile is close to universal access but the unit cost for each in the last disconnected households in remote areas is extremely high due to the challenging geography. In Peru, it has been estimated that the marginal costs for an additional connection will increase exponentially when it reaches about 85 percent water supply coverage. Achieving the Sustainable Development Goal of universal coverage is ambitious and requires more spending, but it will not have the envisioned impact of increasing access rates significantly if the efficiency of capital investment spending is not improved. Spending Efficiency Allocative Efficiency 16. Spending priorities do not appear to be fully aligned with national sector priorities . The National Sanitation Plan aims to (a) modernize WSS sector management; (b) improve the sustainability of WSS services; (c) improve the quality of WSS services; (d) improve the financial viability of the service providers; and (e) improve access to WSS services. While capital spending has increased to improve access to WSS services, the financial viability of service providers has deteriorated (as discussed below) and the quality of WSS services has hardly improved. 17. The budget could play a stronger role in allocating funds to priority areas and close access gaps. The original approved budget for WSS capital spending only shows a weak 97 correlation with the share of the population without improved WSS service (Figure 3-7). Major sources of funding in the sector are recursos ordinarios and canon and royalty payments. Yet, canon and royalty payments are not equally distributed across the country. Some municipalities and regions, which host larger mines, benefit significantly from canon payments, while others do not. Data for 2015 suggest that mid-year budget modifications had a positive impact on channeling more funds to regions with relatively low coverage of WSS services (Figure 3-8). Yet, large budget modifications lead to project cost increases and delays (as discussed below). The introduction of fiscal rules to allocate funds based on sector priorities at the time of budget approval would allow subnational governments to plan and implement their capital budget in line with sector priorities. Figure 3-7 Share of Approved WSS Capital Figure 3-8 Share of Modified WSS Capital Budget and Share in Unserved Population, Budget and Share in Unserved Population, by Region, 2015 by Region, 2015 16% y = 0.5287x + 0.0196 16% 14% 14% Share in Popualtion wihtout Share in Ppulation without R² = 0.3092 imprioved WSS Services 12% improved WSS services 12% 10% 10% 8% 8% 6% 6% 4% 4% y = 0.7262x + 0.0114 2% 2% R² = 0.5514 0% 0% 0% 5% 10% 15% 0% 5% 10% 15% 20% Share in Approved Budget Share in modified budget Source: SIAF and author’s calculations. Budget Execution 18. Large budget modifications lead to under execution of the WSS budget. Large budget revisions are quite common in Peru, but they are exceptionally large in the WSS sector. Between 2009 and 2015, the original WSS budget averaged Soles 23 billion; the final modified budget (PIM) was Soles 36 billion. This 50 percent increase in the budget mid-year is significantly above the national average of 25 percent. Moreover, there are large differences in annual modifications— adjustments range from 36 percent to 92 percent of the approved total WSS budget. This increases the unpredictability of available funds. The situation is even more dramatic for local governments—the modification of the local government budget averaged more than 400 percent between 2010 and 2014 (Table 3-2). As a result of these large and unpredictable budget revisions, actual expenditures generally lag behind modified budget allocations. Regional data shows that the larger the budget modification, the lower budget execution is. In fact, 16 percent of the variance in budget execution rates across regions can be explained by budget modifications alone. 98 Table 3-2 Approved (PIA) and Modified Budget by Type of Government Modified as per 2009 2010 2011 2012 2013 2014 2015 Approved Budget Local 3.03 4.87 3.39 2.96 3.21 4.07 3.45 Regional 4.33 2.61 2.74 2.65 2.42 2.89 2.14 National 0.59 0.47 0.28 0.39 0.32 0.35 0.74 Total 1.92 1.86 1.58 1.69 1.36 1.40 1.76 Source: SIAF and author’s calculations. 19. The ability to spend the budget is especially low for sub-national government entities and capital spending. Execution of current spending as a share of the recurrent adjusted budget (PIM) averaged 85 percent in 2015 for the general government. It was significantly higher for the central government (94 percent) compared to regional or local governments (Figure 3-9). The relative strong execution capacity of the central government can be explained by the fact that its approved current budget tends to be significantly larger than its modified budget after transfers to subnational governments.26 Budget execution of regional governments has improved as their capital budgets have declined. Large disparities in budget execution across regions remain, ranging from less than 50 to more than 90 percent. Local governments struggle to execute their capital budget, with execution rates ranging from 60 to 66 percent between 2009 and 2014. Figure 3-9 WSS Spending in % of PIM at different government levels a. Recurrent Spending b. Capital Spending 100% 100% 100% 100% 80% 80% 80% 80% 60% 60% 60% 60% 40% 40% 40% 40% 20% 20% 20% 20% 0% 0% 0% 0% 2009 2010 2011 2012 2013 2014 2015 2009 2010 2011 2012 2013 2014 2015 Local Regional Local Regional National Recurrent Spending National Capital Spending Source: SIAF and Author’s Calculations. 20. Different rules about the use of funding sources affects capital spending execution. As discussed earlier, the most important funding sources are recursos ordinarios and canons. When recursos ordinarios are not spent, they revert to the Treasury. Canon payments stay with the subnational government; they do not revert back to the Treasury. These unspent amounts become part of the Saldos de Balance, and they can be incorporated into the executed budget at any time. They can also be partially used for recurrent expenditures. Rules about the share of additional canon resources that can be used for recurrent expenditure in maintenance are set every year. 26 In contrast to many other countries, the recurrent budget in the WSS sector includes hardly any staff expenditure but consists largely of project implementation activities (including supervision). 99 Budget execution of recursos ordinarios is significantly higher than that of canon and royalty payments, which is likely the result of their ‘use it or lose it’ nature. 21. The unpredictability of budget approvals leads to significant inefficiencies: i. For multi-annual projects unpredictability of funding can lead to longer implementation periods and higher costs—both in capital costs (as the cost of construction increases because of these delays) and in recurrent costs (as the supervision and implementation support budgets that make up a significant part of the total recurrent costs also increase). ii. It leads to weaker prioritization of projects since it reduces the time to implement projects within a given year (since the modification are only approved mid-year) and creates an incentive to spend the allocation – except for canon payments – on any shovel-ready projects. iii. It can lead to a misallocation of capital investment across municipalities since municipalities that have access to canon payment may not have the largest investment needs. 22. The lack of a sound multi-annual budget framework also creates important inefficiencies.27 Many water sector investments take several years to implement, yet there is no commitment of resources beyond the current budget years, leading to: i. Project fragmentation and higher project costs. Projects are often very small or cut into annual fundable segments, which can result in higher investment costs. The fragmentation causes smaller-size procurement contracts, and reduces opportunities to bundle smaller procurement contracts into large contracts with lower unit costs. ii. Project delays. The annual budget allocation can lead to a stop-go cycle, whereby projects are initiated, stop, and continue depending on annual budget availability. Between 2012 and 2014, close to 5,500–6,000 WSS registered projects were classified as works, of which only about 80–85 percent had a budget assigned to them. 28 The average spending per project is around Soles 580,000 per year—with disbursements ranging from Soles 0 to 165 million—pointing to funding of fragmented and delayed projects. Project delays can lead to significantly higher costs due to modification and renegotiation of contracts. The opportunity costs (forgone returns to investment) of the delayed completion of projects can also be substantial. iii. Overlap and duplication. The municipalities may use different funds for the same project, possibly resulting in overlap and duplication. 27 See also IMF (2015) for an assessment of fiscal transparency in Peru. 28 Only in the data for 2012–2014, the SIAF includes details on works (obras), activities (actividades), and investment actions (acciones de inversion). Obras refer to the improvement, expansion, and/or rehabilitation of WSS infrastructure. 100 Table 3-3 Project Fragmentation Indicator 2012 2013 2014 Total expenditure (million S/) 3,196 3,261 3,449 Number of projects 5,466 5,995 5,874 Number of projects with expenditure in the 4,566 4,892 4,714 current year Share of projects with expenditure 84% 82% 80% Expenditure per project 584,687 543,937 587,113 Source: SIAF and author’s calculations. Efficiency of Investments 23. In general, project selection is sound. The Government manages a public investment system, SNIP. This system ensures, that all investment projects are registered and meet certain criteria. This system uses economic tools to determine the costs and benefits of different investment projects. It is useful in bringing discipline into the capital spending process and having projects systematically vetted and selected based on transparent criteria. Yet, the current appraisal system involves a high cost of compliance, which disadvantages poorer municipalities. 24. Peru’s coverage and investment rates for both, water and sanitation, have led to an increase in marginal costs.29 The average unit cost of increasing or maintaining coverage levels is an important measure of expenditure efficiency. Data for the periods 2003-2007 and 2008-2012 show that the incremental costs of service provision increased six times in Peru over the last decade. This rapid increase may also be due to significant inefficiencies in service delivery resulting from project fragmentation, leading to the implementation of small projects that do not benefit from economies of scale and associated efficiency gains. Hidden subsidies 25. The financial performance of service providers has been weak. Between 2009 and 2014, billed revenues increased from Soles 1.55 billion to Soles 2.21 billion, while O&M costs increased from Soles 1.38 billion to Soles 2.2 billion. Except for SEDAPAL which was one of the few EPS able to generate operating net revenues, most EPS utilities were unable to increase their tariffs in line with the increases they experienced in their O&M cost, generating operating losses (Figure 3-10).30 As a result, the operating cost-coverage ratio (OCCR), which measures to which extent the utility companies’ revenues cover the costs of supplying water and wastewater services, 29 Data constraints limit the assessment of the efficiency of WSS investments. The current data does not allow to disaggregate spending on subsectors (water supply, sanitation, and wastewater treatment) and type of spending (expansion, rehabilitation, and management improvements. The current summary therefore uses data from the Economic Commission for Latin America and the Caribbean (Comisión Económica para América Latina y el Caribe, CEPAL) (2014). 30 There are three categories of WSS service providers. The first are the 50 EPS whose service area covered 17.4 million people in 2014 (compared to 15.3 million in 2009)—almost solely in urban areas. The largest EPS is SEDAPAL, which serves Lima. They serve the majority of urban dwellers. Yet, they are located in areas where the population growth has been around 1.5 percent per year, whereas overall urban population growth stood at 2.9 percent. The second category include smaller providers serving the remaining urban areas, not covered by EPS. The third category of providers serve rural areas. No data are available on the last two categories of service providers. 101 declined for SEDAPAL and other large utility providers (Figure 3-11). The decline in the OCCR for large utility providers is a major concern since larger providers benefit from economies of scale in their operations, unlike many of the small and medium-size providers. Figure 3-10 Operating Loss (S/ millions) Figure 3-11 OCCR in 2014 and 2014 200 2014 150 2013 2012 100 2011 50 2010 2009 - 2009 2010 2011 2012 2013 2014 - 0.50 1.00 1.50 (50) (100) National Small Utilities Medium-sized Utilities Large Utilities Sedapal Other utilities Sedapal Source: SUNASS, author’s calculations. Note: Operating loss is the difference between operating revenues and O&M costs. 26. The weak financial performance of the EPS poses a significant burden on the central government’s budget. The MEF has repeatedly used taxpayers’ money to bail out local water utilities, paying their debts to multilateral and bilateral organizations (by virtue of the sovereign guarantees) and issuing decrees to forgive water utilities’ value added tax debts—collected by utility providers but not transferred to the tax authority. Centrally funded autonomous agencies, under line ministries (MVCS), have financed water and sewerage infrastructure investments in the water utilities’ jurisdictions. These expenses accumulated to US$2 billion during the 1990s and about US$0.5 billion between 2007 and 2010. 27. The weak performance of the EPS also imposes a heavy cost on consumers. The poor performance of the water utilities led to rationing of services in large urban centers in the Coastal areas (e.g., Piura, Lambayeque, La Libertad, Ica). It has pushed citizens to build water storage facilities in their houses, has reverted progress in access to safe water in all regions, and has forced those without access to the network to pay high prices for clean drinking water that they have to purchase from private vendors. As a result, the poor often pay up to 6 times higher prices for drinking water (see above). 28. And it delays progress in achieving universal coverage. The poor performing EPS have insufficient revenues to cover their operating costs, leaving their infrastructure assets without proper maintenance. The lack of appropriate maintenance leads to rapid deterioration of these assets (and subsequent declines in service quality) and the need for costly rehabilitation investments, imposing additional economic costs. As a result, the capital transfers from the MVCS are too often used to replace infrastructure instead expanding the network to improve coverage. 29. The weak financial performance of the utility companies results from a number of factors, including mispricing of water and sewer services, collection inefficiencies, and non- 102 revenue water (NRW) losses. The current tariff regimes do not allow for full cost recovery; in addition, revenue collection from tariffs is inefficient. The tariff regime thus provides implicit subsidies to existing consumers. Moreover, NRW losses provide implicit subsidies to unconnected consumers. All these inefficiencies pose an illegitimate claim on public resources. We quantified these subsidies, using the hidden costs calculator methodology. 31 Overall, hidden costs in 2014 were mostly resulting from underpricing of water and sewer services and excessive NRW losses. 30. The total amount of hidden costs is about US$350 million or 0.2 percent of GDP per year. Underpricing accounts for 79 percent of total hidden subsidies, NRW losses account for 19 percent, and only 2 percent are the result of poor collection efficiency (Figure 3-12). This is not surprising since water tariffs in Peru are among the lowest in Latin America (Figure 3-14). The hidden costs tend to grow over the years, and the underpricing and high losses contribute the most to the total of less than US$350 million per year (which is equivalent to about Soles 1 billion). The majority of these hidden subsidies are linked to Lima’s SEDAPAL, which is by far the largest water utility company in Peru (Figure 3-13). 31 The Hidden Costs Calculator model—developed through collaboration between the World Bank Group and the IMF. (Saavalainen and Ten Berge. 2006. “Quasi-Fiscal Deficits and Energy Conditionality in Selected CIS Countries.” IMF Working Paper, WP/06/43.) It provides insight into three key components of hidden costs: i) tariffs set below cost recovery rates; ii) excessive losses due to inefficient operations; and iii) theft from the networks, which is measured by NRW. The model compares the difference between actual revenues and revenues that could be anticipated in a well-functioning system operating with cost-covering tariffs, bills paid, and losses that are normative for networks of a certain age and design. The methodology of calculations is presented in Annex 2. 103 Table 3-4 Marginal Investment Costs in WSS Coverage (Access Rates and Costs in US$, on PPP Terms and at Constant 2011 Prices, per Additional Connection) a. Access to Improved Water Source b. Access to Improved Sanitation Facilities 2003–2007 2003–2007 Access US$0– US$500– US$1,000 US$1,500 US$2,000 Acces US$0– US$500 US$1,000 US$1,500 US$2,000 > US$5,000 > US$5,000 499 999 1,499 1,999 5,000 s 499 –999 –1,499 –1,999 –5,000 Paragua Bolivia 80– y (161) Peru Nicaragua Ecuador 40– (415) Peru Nicaragua Bolivia(594) (1,393) (1,999) 69% Paragua (557) (2,023) (283) y (156) Guatemala Guatemala Panama 90– (US$1,165) Panama Colombia 70– (1,199) (1,878) Colombia 93% Mexico (1,662) (4,660) 79% Brazil Ecuador (4,089) (1,353) (1,290) (1,621) 94– Costa Brazil Chile 97% Rica (461) (1,279) (3,776) 80– Mexico 89% (1,261) 98– Uruguay 100% (9,267) Costa 90– Rica Chile Uruguay 100% (453) (3,238) (6,233) 104 2008–2012 2008–2012 US$0– US$500– US$1,000 US$1,500 US$2,000 > Access US$0– US$500– US$1,000 US$1,500 US$2,000 > 499 999 –1,499 –1,999 –5,000 US$5,000 Access 499 999 –1,499 –1,999 –5,000 US$5,000 Ecuador 80– Bolivia Nicaragua (3,935) Colombia 90% (570) (1,922) Peru (8,343) 40– Bolivia Nicaragua (3,644) 69% (861) (3,018) El 90– Paraguay Guatemala El Salvador Panama 93% (128) (692) Salvador (2,652) (5,135) 70– Paraguay Guatemala (2,239) Panama 79% (195) (738) (4,722) Peru Colombia (3,348) (7,000) 94– Costa Rica Mexico 97% (1,860) Brazil (2,127) 80– Mexico (2,073) Brazil 89% (1,978) Ecuador (2,157) (3,188) 98– Chile Uruguay 100% (4,702) (6,717) 90– Costa Rica Chile Uruguay 100% (1867) (3,868) (6,691) Source: World Bank staff estimates based on CEPAL 2014. World Development Indicators data. 105 Figure 3-12 Hidden subsidies (S/ millions) Figure 3-13 Investments (S/ millions) 400 350 300 38% 250 200 54% Millions USD 150 6% 100 2% 50 0 Sedapal Small companies 2011 2012 2013 2014 Medium companies Large companies Underpricing Under collection Due to high losses Source: SUNASS, author’s calculations. Figure 3-14 Average Residential Water Rates in Latin America and the Caribbean per m3 32 Source: SUNASS, International Benchmarking Network for Water and Sanitation Utilities (IBNET), and author’s calculations. 32 Based on a Consumption of 15 m3 per Connection per Month. 106 31. Hidden costs—if unaddressed—are likely to reduce economic lifetime of the investments and the quality of service provision. Utilities tend to compensate for hidden costs by reducing or delaying repairs and maintenance; reducing hours of operation: or reducing other aspects of the quality of services. They can trigger a downward spiral of significant deterioration in the value of WSS assets, declining service quality, and increasing cost for each unit of WSS service provided. As a result, losses increase substantially and abnormally high investments are required for rehabilitation, leading to further inefficiencies and higher public expenditures. 32. The link between investment planning and tariff increases has worsened. According to current legislation in place since 1994, investment programs within the water utilities’ service area are to be approved by their Board of Directors. After these programs are approved, the investment program must be discussed with SUNASS to determine the tariff implications. Tariff increases would need to be linked to progress in the implementation of investment programs. Also, if the program has financing from the central government, it would need to be authorized by the Directorate of Multi-Annual Programming, at MEF. In practice, investment activities occur outside of the current legislation. If investments are made without approval of the commercial governance structure of the water utilities and SUNASS, tariff increases to operate and maintain the new infrastructure may not be granted. The costs of moving towards universal coverage under different efficiency scenarios 33. Moving towards universal coverage in water and sanitation by 2021 without realizing efficiency gains in spending has been estimated to cost at least 20.3 billion Soles or about 0.6 percent of GDP annually. It has been estimated that these accumulated public expenditures are required to bring access to water to near universal levels and to increase access to improved sanitation to 92 percent nationally and 50 percent in rural areas. The costs to improve access to improved sanitation account for three-quarter of the total costs. The estimated costs do not take into account the potential costs of additional investments in rehabilitation and other complementary investments. It is estimated that these additional rehabilitation and complementary investments costs could amount to 13.5 billion (an additional 0.4 percent of GDP each year from 2017-2021), bringing the annual costs to achieve universal access to almost 1 percent of GDP for the next five years. 34. The costs would be significantly lower if the government would implement reforms to boost the efficiency in public expenditures in the water and sanitation sector to the levels observed in other countries in the region with similar topographic challenges and higher coverage rates. Brazil and Mexico have to some extent similar topographies as Peru with either large areas covered by Amazon rainforest or a significant share of the population living in high altitude mountain ranges. Both countries have already achieved near universal access rates of about 94-97 percent by 2012 (see Table 3-4). Nevertheless, both countries have significantly lower marginal costs in raising the access to water and sanitation to more households. It is estimated that the accumulated cost of achieving universal coverage would decline by more than half (51 percent) if Peru would achieve the same level of public spending efficiency in the water and sanitation 107 sector as Brazil.33 Achieving Mexico’s efficiency levels would generate accumulated savings of up to 0.3 percent of GDP annually between 2017-2021. Table 3-5 Policy options and potential savings Cost of coverage for water and sanitation (Million soles of 2016) Scenario 4 Scenario 1 Scenario 2 Scenario 3 (Mexico (Base) (Base -20%) (Brazil cost) cost) Water Supply Urban 2,859 2,287 3,072 3,029 Rural 2,244 1,796 1,035 1,021 Total 5,103 4,083 4,107 4,050 Sanitation Urban 10,073 8,058 4,325 4,127 Rural 5,147 4,118 1,597 1,524 Total 15,220 12,176 5,923 5,651 Total WSS 20,323 16,258 10,030 9,701 Price assumptions: price per connection in 2016 soles Scenario 4 Scenario 1 Scenario 2 Scenario 3 (Mexico (Base) (Base -20%) (Brazil cost) cost) Water Supply Urban 3,604 2,883 3,873 3,819 Rural 8,395 6,716 3,873 3,819 Average 4,812 3,849 3,873 3,819 Sanitation Urban 8,667 6,934 3,722 3,551 Rural 11,993 9,594 3,722 3,551 Average 9,564 7,651 3,722 3,551 Source: Author’s calculations. 33 The following assumptions have been made: (i) the same marginal costs are used for urban and rural areas in Brazil and Mexico as more detailed cost estimation were not available, (ii) prices have been updated to 2016 soles, (iii) we have not considered additional investments in rehabilitation and other complementary investments, which may be important, iv) we are only assuming urban universal coverage (100 percent) for water and sanitation and rural partial coverage: 92 percent for water and 50 percent for sanitation. 108 Conclusions and Policy Options 35. Large inefficiencies in the sector have limited the progress in increasing access to improved WSS services. The significant increases in spending that started in 2005 has not yet resulted in the anticipated progress, outlined in the National Sanitation Plan. Improving the efficiency with which resources are used would do much to improve the absorptive capacity of the sector, increase the efficiency of public expenditure, and ensure improvements in access to and the quality of services. To that extend, the following measures are suggested: (a) improve sector investment planning; (b) improve the incentive structure in budget and investment allocations; and (c) improve capital and operational efficiency, including the use of water pricing as a tool to improve efficiency while safeguarding the affordability of access. As the sector is highly decentralized, this will require strong coordination between the different stakeholders in the sector. 36. Preparing a medium-term investment strategy for the water sector, that includes a transparent procedure to prioritize investments over a multiyear planning period, will help improve the predictability of WSS sector funding. The investment plan will help the sector move away from the currently highly discretionary mechanisms and, instead, set medium-term budget ceilings for municipalities that help them to better plan investments. Multiannual budgeting will also help municipalities to account for their annualized costs over the economic life of their assets. Collaboration and coordination is key to ensure that such a strategy can be prepared and implemented. The multiannual investment plans should be directly linked to the budget and hence should include not only the prioritized investments but also the financing of these investments. The current assessment of projects takes place mostly in isolation. Each project is being assessed individually, but not in the larger context of the total investment needs. A recent World Bank study (2015)34 for SEDAPAL shows that utilities would be able to save 25 percent of their investment costs (equivalent to about 2 percent of GDP), without compromising their services, if they would evaluate their total investment plans in a more comprehensive manner. Since a large part of the O&M costs are the result of past investment decisions, inefficiencies in investment planning lock in O&M costs for decades. 37. Better coordination and collaboration between the different levels of government is key to ensure that investment decisions improve and reduce future O&M costs. The fragmentation of projects is a real problem as it results in a large amount of very small projects that are costly to administer and whose impact is relatively small. The small size of many projects is also linked to the late arrival of central government funds, which results in ad hoc investment decisions to speed up project execution rates towards the end of the year. 38. Policymakers should improve the WSS budget rules to align the sector objectives with the investment programs and effectively target the municipalities and their WSS investment and technical assistance requirements. Budget allocation rules and expenditure management guidelines should be prepared and used during project budget preparation, reducing the need for budget modifications during budget implementation. These rules can be periodically revised, but should be relatively stable to allow predictability in fund allocation. The rules should also ensure 34 Kalra, Nidhi Rajiv; Groves, David G.; Bonzanigo, Laura; Perez, Edmundo Molina; Ramos Taipe, Cayo Leonidas; Cabanillas, Ivan Rodriguez; Brandon, Carter J. 2015. Peru - Robust decision-making in the water sector: a strategy for implementing Lima’s long-term water resources master plan. Washington, DC: World Bank. 109 that municipalities with different sources of funding are treated equally, as the ‘use it or lose it’ nature of some of the funding sources may provide perverse incentives to spend at all costs. 39. Introduce, where possible, more performance-based incentives in the transfer of funds. A large part of the sector’s funding is being transferred to service providers while the performance of these providers leave ample room for improvements that can reduce the cost of the service and increase the revenue flows. It would be useful to provide these funds to service providers in a way that links investments, especially expansion investments, and compliance to a minimum performance standard to ensure that spending affects the outcomes that the Government wants to achieve. An example would be the provision of expansion investments only when certain levels of performance are achieved; the use of results-based contracting and financing, or other allocation incentives based on verifiable operating improvements. In rural service provision, a least subsidy approach could be used to allow for a more efficient use of resources. 40. Water and sewer tariffs are low in Peru (especially compared to many of its regional peers) and are a major reason for the widespread existence of financially unsustainable service providers jeopardizing the sustainability of infrastructure investments. Hence, there is space for tariff increases. Before making any changes to the tariffs, utilities should first address their low billing and collection efficiencies. Fixing gaps in the utilities’ billing and collection systems will generate additional revenue and reduce the size of tariff increases that will ensure the financial sustainability of the water supply service. The sector should also start to think about rebalancing the tariff structure to ensure that it includes discriminatory tariff policies whereby poor consumers pay less (preferably if possible by the use of direct subsidies), whereas those that can pay do so, while also broadening the customer base where possible. The current use of cross- subsidies may result in some nonresidential users moving out of the system; this may require the pricing of alternative water sources, such as groundwater, to ensure that the price differential between piped water and alternative sources does not provide incentives for larger and nonresidential users to leave the EPS for their water and sewerage services. 41. To further improve the efficiency of investments, the Government may rethink some of its standards and norms. In view of large investment cost differentials, the Government can put in place a wider range of technologies (that allow for a set of minimum standards to be met) or use variations in budget allocations to cope with more challenging conditions (which will also require more transparent subsidy policies). Over the longer term, policy makers will need to develop a more thorough understanding of investment and operational costs in the sector to realign investment incentives and ensure the long-term financial sustainability of the WSS sector. 110 Table 3-6 Policy options and potential savings Policy Context Action Efficiency Equity Potential Comments Savings (% of GDP) Levels of cost Reduce + Neutral 0.132 When utilities do not price their recovery in the underpricing of services properly, will postpone water sector are water and maintenance (which will affect low wastewater service quality and payment services for EPS. behavior of customers) and in the longer term, will result in a large program of rehabilitation investments, which are financed by the government. Reduce collection + + 0.003 efficiencies of EPS. The high and Reduce NRW + ? 0.032 This may be an overestimation as increasing leves losses in the sector the reduction of these losses may of NRW serviced by EPS. require upfront investment costs. The Improve + + 0.13–0- A recent study by the World Bank fragmentation of investment .18 showed that SEDAPAL could save WSS investments planning and 22 percent of total investment costs heightens risks of coordination at and if it used a more flexible approach duplication, between different to planning and prioritization of redundancies, and levels of investments that looked at how the financial leakage. government. combination of a set of different Investment investments could achieve the long- projects are term goals. prepared and Assess individual + + n.a. By reducing the overall investment implemented by project viability as costs, the cost of project preparation different part of the larger and supervision can be reduced. ministries, investment regional and local program, which governments, and can help reduce the other public transaction costs agencies for the preparation (including EPS, and FONAFE, and so implementation of on) without much projects. coordination Improve incentive + ? n.a. Reduction of investment delays and between the structure in the related costs (for example, cost of different budget process by supervision) agencies. (a) decoupling the budgetary and investment process to allow for multiannual investment planning and budgeting and (b) changing rules that create adverse incentives. 111 Policy Context Action Efficiency Equity Potential Comments Savings (% of GDP) More transparent + + n.a. The current allocation of funding is formulas to assign highly discretionary and results in funding highly unpredictable funding that generates costs in terms of implementation delays. More detailed data are needed to determine the costs of these delays. Note “n.a.” = Not applicable 112 Introduction 1. Peru has been able to improve its health indicators in recent years, but the sector still faces various challenges. The cost paid by the households is relatively high compared to other OECD countries, but it is aligned to the expectations in view of the country’s per capita income. Around 45.5 percent of the vulnerable urban population is covered by the Integral Health System (SIS). The big challenge for the provision of public services should be seen within the framework of the demographic and epidemiological transitions that go hand in hand with the urbanization and development process. The demographic transition is giving rise to a higher average life expectancy and to ageing of the population. Some conditions, related to neuropsychiatry, malign tumors and cardiovascular problems, are appearing within the epidemiological transition, generating expenses that will create additional pressure on the sector’s fiscal cost in the future. 2. Public health expenditures show significant inefficiencies; removing these inefficiencies can generate fiscal savings that can be used to address the increasing pressure to improve services in the sector over the medium-term. There are four areas of inefficiencies: the fragmented architecture of the system, the limited prioritization of expenditures, the mechanisms to finance suppliers, and the lack of systems to monitor expenditures, services and results. Addressing these inefficiencies requires investments in modernizing the provision of healthcare and the public health information systems in order to plan the budget and execute the expenditure more efficiently. In particular, we estimate that a reform of the system’s fragmented architecture system may generate efficiency gains of up to 0.3 percent of GDP. More specifically, eliminating the cross-subsidies among the subsystems that lack transparency may generate a gain of 0.1 percent of GDP. The reduction of the health system administrative expenses to the level of peer countries, for example by eliminating duplications in the administration and management among the different fragmented systems, may generate efficiency gains of between 0.1 and 0.2 percent of GDP. Moreover, this chapter presents other reform options aiming to improve efficiency of the sector and possibly generating savings in the medium term, but which are by their very nature more difficult to calculate. 3. The objective of this chapter is to identify the most important sources of inefficiency in Peru’s health system. Based on an analysis of the roles of the institutions and the financing instruments supporting the health insurance framework, recommendations are issued for the different actors, such as MINSA, the Regions, the Integral Health System (SIS), MEF, SUSALUD, EsSalud and SUNAT. Expenditure and Results 4. Total expenditures on health relative to the Gross Domestic Product (GDP) have grown, primarily due to an increase in public spending. The total spending in health in terms of GDP increased from 4.8 percent in 2000 to 5.5 percent in 2014. Private spending remained stable (from 2.1 percent of GDP in 2000 to 2.2 percent of GDP in 2014), while public spending (excluding Social Security) increased from 1.3 percent of GDP in 2000 to 2.3 percent of GDP in 2016 (see Figures 4-1 and 4-2). The trends in the subsectors varied: aside from social security, spending increased from 1.3 percent to 2.3 percent of GDP between 2000 and 2014, while the 113 health expenditure on Social Security fell from 1.4 percent to 1.0 percent of GDP. The health expenditure (excluding Social Security) as a percentage of total public spending increased from 7.7 percent to 10.8 percent between 2010 and 2016 (see Figure 4-2). On the other hand, the out- of-pocket expenditure on health displays a significant decrease from 41.4 percent to 28.6 percent between 2008 and 2014.35 Figure 4-1 Total health spending, % of Figure 4-2 Health spending between 2009 and GDP 2016 6% 5.5% 2.6% 12% 4.8% 2.4% 2.3% 2.4% 2.2% 11% 4% 2.1% 2.2% 10.8% 3.3% 2.0% 1.8% 10% 2.7% 1.0% 1.8% 9% 1.6% 2% 1.4% 1.6% 8.9% 2.3% 8% 1.4% 7.7% 1.3% 7% 0% 1.2% 1.0% 6% 2013 2014 2011 2003 2007 2005 2010 2004 2012 2001 2000 2008 2002 2006 2009 2009 2010 2011 2012 2013 2014 2015 2016 Private Public Soc. Sec. Public Public, incl. Soc. Sec. Spending in health (% of GDP) Total Spending in health (% of total spending) Source: World Bank staff based on SIAF (Public Source: World Bank staff based on WB WDI (2017-01-31). spending excl. Social Security), WB WDI 2017-01-31 Note: Public spending in health (MINSA, GORE and Social (Public spending including Social Security) and WB Security), % of public spending. WDI 2016-09 (Private Spending). 5. The level of total health expenditures relative to GDP is lower than in peer countries in the region but comparable to the spending levels of structural peers outside the region. The total expenditure on health relative to GDP in 2014 was 5.5 percent, while the average in Latin America was 7.2 percent and the average of the Organization for Economic Cooperation and Development countries (OECD) was 12.4 percent (see Figure 4-3). However, the expenditure is in line with structural peers outside the region and is commensurate to the level of revenues. Despite the recent decline, out-of-pocket expenses are still relatively high compared to OECD countries (14 percent). Still, they are also commensurate to the country’s income level (see Figure 4-4). 35 Source: WB WDI. It should be noted that the percentage estimates of out-of-pocket costs in the WB WDI were revised downward compared to previous estimates. 114 Figure 4-3 International comparison of the total spending in health, % of GDP, 2014 18% Total spending in health, % of GDP, 2014 16% 14% 12% Canada 10% Thailand Chile Australia Colombia 8% 6% Peru Poland 4% Rumania Malaysia 2% 0% 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000 100 000 GDP per capita, 2014, PPP,$ current international Peru Structural peers Regional peers OECD average OECD Source: World Bank based on WB WDI (2017-02-01) Figure 4-4 International comparison of out-of-pocket expenses in health, 2014 45 % Out-of-pocket expenses, % of total 40 % expenditure on health, 2014 Malaysia 35 % Peru Chile 30 % 25 % Poland Australia 20 % Rumania 15 % Colombia 10 % Canada Thailand 5% 0% 0 10 000 20 000 30 000 40 000 50 000 60 000 70 000 80 000 90 000 100 000 GDP per capita, 2014, PPP,$ current international Peru Structural peers Regional peers OECD average OECD Source: World Bank based on WB WDI (2017-02-01) 6. Peru has made considerable progress in improving its health indicators, achieving the target of the Millennium Development Goals (MDGs) by reducing the infant mortality rates by two thirds between 1990 and 2015. According to WB WDI, in 2015 the national infant mortality rate was 13.1 deaths per 1,000 live births, compared to 56.3 in 1990 (see Figure 4-5). Likewise, the prevalence of chronic child malnutrition in children under 5 years old declined from 115 31.3 percent in 2000 to 14.6 percent in 2015. 36 The reduction of child malnutrition is explained, on the one hand, by the economic growth and, on the other hand, by the application of a multi- sectoral public strategy that combined financing based on nutrition targets (results-based budgeting), adopting a multi-sectoral territorial approach, coordinating interventions at the local level, and strengthening the demand for health services through the Joint Conditioned Transfers program. Figure 4-5 Infant and under-5 mortality rate 90 80 79.7 70 60 Infant mortality rate (per 1,000 56.3 50 live births) 40 Under-5 mortality rate (per 30 16.9 1,000 live births) 20 10 13.1 0 1985 1990 1995 2000 2005 2010 2015 2020 Source: World Bank based on WB WDI (date of download: 2017-02-01). Note: The infant mortality rate is the probability of dying between birth and the age of 1, per 1,000 live births in Peru, 1990-2015. 7. Despite these achievements, Peru is still facing various challenges in maternal and child health. First, the prevalence of malnutrition is still high in rural areas, the Amazon forest and the highlands (INEI 2017). An even more alarming element is the persistence of anemia. According to the Population and Household Health Survey (ENDES) of 2015, 32.6 percent of children under the age of 5 suffered from anemia, which was higher than in 2011 (30.7 percent). The percentage of children with anemia was even higher in rural areas (39.7 percent) and in some departments, such as Puno (61.8 percent), Pasco (45.9 percent), Madre de Dios (44.3 percent) and Loreto (44.0 percent) (INEI 2016). The World Health Organization (WHO) describes a prevalence of anemia above 40 percent as a serious public health issue. According to the international NGO Action Against Hunger, the costs resulting from anemia are estimated to be around US$ 650 billion, approximately 0.55 percent of the country’s GDP (Action Against Hunger 2012). Peru’s health system: Vision and challenges 8. The country aspires to achieve universal health insurance. This aspiration is not unique: the international context shows that the challenges which emerging countries face is to move forward on the path of replacing financing based on out-of-pocket payments by prepay or insurance-based financing. By prepaying health expenses, it is possible to group the insured’s contributions and share the risk of having to pay for medical care among all members of the fund. Starting with Free School Insurance and Maternal and Child Insurance in the second half of the 1990s, Peru has sought to gradually broaden insurance coverage in terms of population and health 36 According to INEI, in 2015 the national infant mortality rate was 15 deaths per 1,000 live births, compared to 55 in 1991/1992. INEI estimates that the percentage of children under 5 years old with chronic malnutrition fell from 36.5 percent in 1991/1992 to 14.4 percent in 2015. INEI 2017: Perú: Evolución de los indicadores de los Objetivos de Desarrollo del Milenio al 2015. http://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1413/libro.pdf. 116 needs. In 2002 the two insurance schemes merged into the Integral Health System which has progressively integrated poor and vulnerable population groups besides from enhancing coverage of the number of health conditions. There is also EsSalud, a health insurance scheme for the population employed in the formal sector of the economy as well as their families. The Framework Law of Universal Health Insurance (AUS) (2009) formally confirms the country’s aspiration to have a universal health insurance which is defined as “a process aimed to achieve that the entire population living in the national territory would have access to health insurance and healthcare services in the field of prevention, promotion, recovery and rehabilitation in adequate conditions in terms of efficiency, equity, opportunity, quality and dignity, based on the Essential Health Insurance Plan (PEAS).” 9. The AUS Law provides the legal framework to achieve the objective of the universal health insurance and does not consider a radical change in the structure of the system. The multiple Institutions Administering Health Insurance Funds (IAFAS) continue to exist, including the public insurance scheme (SIS), the private insurance and social security schemes (EsSalud and Sanidades de las Fuerzas Armadas y Policiales (health funds of the armed forces and police)). No changes are made either on the level of the multiple providers, including public healthcare providers (MINSA, Regional Governments), private players, EsSalud and the health fund of the armed forces. 10. The AUS Law aims to improve the links among the different agents through three lines of action: i. The Law refers to three insurance schemes: people’s contributions to an insurance scheme on their own account or through their employer (essentially the formal sector); the subsidized scheme targeting the most vulnerable population groups through the Integral Health System; and the scheme based on partial contributions with partial public financing. ii. The Law provides for a minimum package of benefits, called the Essential Health Insurance Plan, which is mandatory for all IAFAS. iii. The Law proposes to link health service providers through: (a) the standardization of processes and procedures, and (b) the development of mechanisms for payment and the exchange of services that enable purchasing and selling services. It is therefore clear that the Law strives to reduce the health care system’s fragmentation through interoperation among insurance providers and healthcare providers. 11. Peru went through a radical change in terms of the composition of its population between 1997 and 2015. These changes affected the share of the population being poor or vulnerable and the proportion of the population working in the formal or informal sector. While the poverty dimension is important because it is linked to the ability to pay for insurance and services, belonging to the formal sector is important because in Peru formal employment is equivalent to coverage under the social security scheme (EsSalud). Combining both dimensions, we define four population groups: the poor population (according to the definition of the National Statistics and Information Institute – INEI), the vulnerable or quasi-poor population (with a spending level above the poverty line but at risk of falling into poverty of more than 10 percent), the non-vulnerable population classified in the informal employment sector, and the non- 117 vulnerable population working in the formal sector. Since 1997 the percentage of poor and non- vulnerable populations have reversed: the poor population fell from 55 percent to 24 percent in 2015, while the non-vulnerable population increased from 27 percent to 54 percent (see Figure 4- 6). In 2015, the informal non-vulnerable population accounted for 24 percent of the population. Although this population group is not poor and is not at risk of falling into poverty, they do not have access to health insurance through social security since none of their household members are employed in the formal sector. The size of this population group is almost equal to that of the poor and vulnerable. Figure 4-6 Population (1997 vs. 2015) Population 1997 Population 2015 Non-vulnerable, Formal At least one household member works in the formal sector Non-vulnerable, Informal Not one household member works 27 % in the formal sector 30 % Vulnerable Per capita household spending 18 % above the poverty line but below 24 % 1.4 times the poverty line 55 % 24 % Poor Per capita household spending 22 % below the poverty line (INEI) Source: WB. 12. Health insurance coverage expanded rapidly between 2006 and 2015. According to ENAHO, the percentage of the population with insurance increased from 38 percent to 73 percent. In addition, the insurance expansion was higher among the poor and vulnerable: insurance coverage for the poor increased from 32 percent to 79 percent in those years (see Figure 4-7). The increase in coverage can be explained by the growing coverage through SIS and EsSalud. Private insurance coverage has increased as well, although there is no evidence that the coverage of SIS and EsSalud has expanded at the expense of private insurance. Our analysis has shown that SIS coverage differs by degree of poverty and vulnerability: Although SIS has been very effective to broaden targeted coverage under an insurance, especially in the rural areas, at the moment the new challenges in terms of insurance coverage are located in the urban area. SIS only covers around 45 percent of the vulnerable urban population, while there is a decreasing coverage of this population under EsSalud, possibly due to a substitution between both insurance institutions. 118 Figure 4-7 Health insurance (2006-2015) 90 % 80 % % of the population covered by 70 % 60 % Poor health insurance 50 % 40 % Vulnerable 30 % Non-vulnerable 20 % 10 % 0% 2005 2007 2009 2011 2013 2015 Source: WB. 13. Nonetheless, seven years after enactment of the AUS Law, a series of important constraints have become apparent in implementing the necessary process to achieve universal insurance. It has not been possible to achieve a significant expansion of the scheme based on partial contributions; the package of benefits of the PEAS is not being used to prioritize services, investments, or to monitor effective coverage of the population; and the exchange of services (based on purchase and sale) between EsSalud and the public system is virtually inexistent in practice. Evidence points to a continuous gap in the (perception of the) availability and quality of services between EsSalud and the public scheme (Raffo 2015, see Figure 4-8). Figure 4-8 Service gap: MINSA/GORE subsystem and other subsystems 160 135 35% 30% 140 30% 120 % of patients 80 Minutes 100 25% 80 64 57 16% 20% 60 15% 40 20 10% 5% 0 5% 0% 0% Delay between arrival at IPRESS and start of the Percentage of patients who paid for all medication out medical consultation, 2016 of their pocket, 2016 Source: SUSALUD 2016. Note: Delay between arrival at the Institution Providing Health Services (IPRESS) and the start of the medical consultation, 2016; percentage of interviewed patients who paid for all medication out of their pocket, 2016. 14. ENAHO data shows that effective access to health for the poor and vulnerable is a still challenging. The poor and vulnerable in Peru have a higher probability of not using health services if they involve out-of-pocket expenses. For example, based on the 2014 ENAHO data, we calculated that 65 percent of the non-vulnerable population that had an illness or accident in the 119 last four weeks prior to the survey received medical care without any out-of-pocket expenses; among the poor, this is only 37 percent. Furthermore, there are major differences in terms of the availability of health services by geographical area: in general, there are scarce human resources, specifically in areas where poor people insured under the SIS live (World Bank 2016). 15. The ability to deliver quality health services is directly related to the population’s satisfaction. The national user satisfaction surveys conducted by SUSALUD (SUSALUD 2016) and a study conducted in 2015 (Raffo 2015) agree that the most important complaints of the health system users are related to waiting times to get an appointment and in admissions, and the lack of medicines covered by the insurance. Moreover, the study shows that the separation in service provisions between the MINSA and EsSalud, and the obvious differences in care and quality, give rise to dissatisfaction among public system users. Efficiency of the Expenditure on Health The Health System Fragmented Architecture 16. The different service delivery systems of EsSalud, MINSA, the Regional Governments, the Armed Forces, and the Police, operate in a parallel and uncoordinated manner leaving gaps in the access to the services and with duplications in infrastructure and administration. For example, most outpatient medical centers belong to the GORE/MINSA network or are private. In contrast, EsSalud has few (approximately 300) outpatient care centers outside its hospitals. While the public sector care network comprises a large number of primary care centers in rural areas, EsSalud has a more focused care network in urban hospitals (see Figure 4-9). The existence of different types of services suggests that these different networks may complement each other in order to deliver a more complete and locally accessible package to the population. In this sense, the Universal Insurance Framework Law (Law No. 29344, 2009) and its Regulation (SD No. 008-2010-SA) provides for the “exchange of services” among the different IAFAS and networks. 17. Despite the significant efforts made in recent years to encourage the exchange of services between EsSalud and the public system, the interoperation between both systems is practically inexistent. The lack of progress can be explained by various reasons, including a scarce agreement about how to fund health services, a limited compatibility among the systems for the purchase of services used in different sectors, a saturation of hospital services in EsSalud, the lack of billing mechanisms or incentives in the primary healthcare units of the public sector and the lack of debt settlement mechanisms among regional governments, municipalities and EsSalud. 18. There are important cross subsidies among the subsystems amounting to an estimated 7 percent of the public budget in health or 0.13 percent of GDP. For example, the “hidden” subsidy of the public sector towards EsSalud is generated when the latter’s insured are “transferred” to public providers. In these cases, the people ensured at EsSalud have to make a “co-payment” when receiving the service. Nonetheless, this co-payment covers only part of the service delivery costs, since the largest part is financed with Treasury resources. At the same time, people covered by SIS and people who are not insured do not have access to healthcare centers of the EsSalud network except in emergencies. Estimates are that the public system’s hidden 120 subsidies towards the people covered under EsSalud may amount to up to 850 million soles per year, i.e., 7 percent of the public health budget or 0.13 percent of GDP. Unlike what may happen in other countries, the state does not bill EsSalud for providing services to its insured. Therefore, it is no surprise that EsSalud has not invested in expanding its primary health services outside of hospitals and has no incentives to move forward with a true “exchange of services”. As long as this subsidy is not made transparent, EsSalud will not invest in primary healthcare and will not grant access to the population covered by SIS to its third-level hospitals. 19. According to our estimates, a reduction in health administration costs to the level of peer countries may give rise to efficiency gains of between 0.1 and 0.2 percent of GDP. The health system fragmented architecture creates duplications in administration and management across the different fragmented systems, resulting in administrative high costs. The administrative expenditure on health in Peru explains 15 percent of the total current expenditure in the sector. This is higher than the level in Mexico (9 percent) or the average of the OECD countries (3 percent). If Peru were able to reach the same level of administrative expenditure as Mexico (in percentage of current expenditure on health), it would save 0.1 percent of GDP; while the same level of administrative expenditure as the average of the OECD countries would save 0.2 percent of GDP. Figure 4-9 Number of outpatient and inpatient (hospitalization) centers, by network of providers Outpatient centers Inpatient centers 8000 350 7000 300 6000 250 5000 200 4000 150 3000 2000 100 1000 50 0 0 Centers, posts and Specialized, diagnostic General Specialized Health centers doctor's offices and dental services hospitals hospitals with beds EsSalud GORE/MINSA EsSalud GORE/MINSA Private Armed Forces/Police Private Armed Forces/Police Subregional governments Others Subregional governments Others Source: World Bank (2016) based on the National Registry of Health Centers (RENAES). 20. The current major challenges pose the question if it would be possible to sustain the universal insurance with the current health system architecture or if there is a need for a transition towards the unification of the system. Historically, it was thought that the big spending difference between EsSalud and the public sector was an impediment in this regard. Nonetheless, the current reality shows that the spending gap between EsSalud and the public sector has become dramatically narrower in the last decade. While in 2005 the expenditure per person in EsSalud was 3.5 times higher than the expenditure in the public sector, in 2014 this had decreased 121 to 1.5 (see Figure 4-10). Likewise, the figures show a major disparity in funding among the regions, both with regard to EsSalud and SIS; although, in certain regions, the difference in the spending per person between EsSalud and SIS is very small (World Bank 2016). Many OECD countries have taken advantage of similar circumstances to reduce their health systems’ fragmentation. Estimates are that the joint financing gap was approximately S/5,518 million in 2014, equivalent to approximately 46 percent of the public expenditure of the MINSA/GORE system (S/12.1 billion in 2014). Figure 4-10 Per person spending of the EsSalud and MINSA/GORE/SIS systems 1000 905 770 EsSalud Per person spending, 800 724 contributors constant S/. 2014 600 516 400 Av. number of 520 SIS non-contrib. 200 260 and contrib. 182 180 0 1995 1997 2005 2014 Source: World Bank 2016. The prioritization of expenditure to promote equity and timely access 21. The prioritization of spending should be based on the levels of care, highlighting primary healthcare and prevention. Even though results-based budgeting emphasizes various lines related to primary healthcare, in general, the budgetary process does not seem to give due attention to this guideline. Moreover, SIAF expenditures records do not distinguish the level of care and so it is not possible to identify or monitor the expenditure executed in the different levels using this database. 22. Even though the AUS Law provided a basic package of benefits with the Essential Health Insurance Plan (PEAS) as the mechanism to prioritize expenditures, this instrument is not used effectively in planning, prioritization, and monitoring of spending. With regard to the prioritization, the SIS has expanded its package of benefits and has created the Intangible Solidarity-based Health Fund (FISSAL) for high-cost illnesses. Although it is true that this complementary package responds to certain needs of the population, this decision was not clear with regard to its service delivery capacity (either of the basic package or the complementary package), the effectively covered population or the packages’ provision costs. In particular, the AUS Law provided for a two-yearly update of the PEAS package based on actuarial studies, which has not happened from 2009 to date. The differences found in this area in part come from the AUS Law itself, which states that the Institutions Administering Health Insurance Funds (IAFAS) can adopt complementary service packages and that once the benefits have been acquired, the insured will enjoy irreversibility, in the sense that the benefits cannot be cut back. For the private IAFAS, the decision to broaden a package of benefits has an impact on its economic profitability. Nonetheless, for the public IAFAS, the decision to broaden a package of benefits has fiscal implications that are not transparent. With regard to implementation, no statistics have been published on the degree of delivery of PEAS in various population groups, or on the public 122 expenditure on services pertaining to PEAS. Considering the existing context with a limited supply and financing, the decision to expand the package of benefits has given rise to major inequities in the access to services, which are rationed because of supply-side elements and not because of their prioritization. A recent study has shown that patients pay private sector doctors to have a preferential access to services in the health system (Raffo 2015). 23. One of the critical functions of health insurance is the identification of the beneficiaries, which, in theory, enables the prioritization of expenditures. Individual identification avoids duplicity between insurance companies in the contributory and non- contributory systems and is helpful to monitor and solve the insurance gap. Substantial progress has been made in Peru identifying the population covered by health insurance schemes, thanks to the creation of the Register of People Covered by the Universal Health Insurance, and the creation of the National Health Superintendency (SUSALUD) as the regulator and administrator of the Register and the link thereof with the National Register of Identity and Civil Status (RENIEC). Nonetheless, there are still major discrepancies in terms of the coverage due to a lack of a standardized definition of what an insured person means. Both in the SIS and in EsSalud, the numbers on coverage overestimate the effective rate of insurance. With regard to the SIS, there are around 3 million insured who are unaware of their rights as a result of the massive and automatic registration processes that were used (see Figure 4-11). Figure 4-11 Discrepancy: SIS coverage rate, by source of information, 2014 90 % 80 % 70 % 60 % 50 % 40 % 30 % 20 % 10 % 0% ENAHO data Administrative data SUSALUD data Source: World Bank 2016 based on ENAHO, SIS administrative data, SUSALUD administrative data. 24. Unlike other countries, Peru does not use the Universal Insurance Register to allocate funds. For now, this Register is used in a limited manner for monitoring coverage. In contrast, in other countries the insurance register is used to facilitate the collection of contributions to the health insurance, to limit the filtration of subsidies to the social security, to avoid overlaps with 123 other social programs and to facilitate the billing of services granted by public or private providers to the insured population. 25. In EsSalud, it is not clear what mechanism is used to prioritize health expenditures. No package of benefits is being used. The current legislation regarding insurance does not allow for exclusions or limitations in terms of care for prevention or recovery. Even though it was not possible to break down the expenditure of EsSalud by level of care, the limited emphasis this institution has put on having first-level centers suggests that the expenditure has an excessive focus on hospital care for recovery and emergency care rather than on prevention and primary healthcare. Just like in the public sector, this is also generating rationing because of the supply-side constraints. The mechanisms for the purchase of services 26. The limited progress made by the SIS as a purchaser of services is another major challenge. The mechanisms to pay for or buy services determine the extent to which it is possible to obtain more and better-quality services with a certain budget. 37 The SIS tried to modernize its systems to pay providers by introducing the so-called “capitation” payment modality in primary healthcare and a service-based payment modality in the hospitals. Nonetheless, the SIS finances only 12 of the costs of the services delivered to its insured, while the rest is still financed from other sources, basically the historical public budget, funds allocated through the results-based budgeting (RBB) and out-of-pocket expenses. 27. The planning and execution mechanisms of both the RBB and the SIS seek to improve efficiency by linking result targets to budgets or payments—but reviewing the mechanisms used in practice shows that most payments are still input- and not result-based.  With regard to the RBB, there is evidence suggesting that despite the major efforts made during the planning phase, in practice payment is not results-based but input-based. The data on budget execution do not reflect the costs of programs or the cost of achieving the goals of the budgetary programs. In practice, the Execution Units charge the expenses to satisfy multiple needs to various sources of financing. Therefore, the amount recorded in the budgetary programs does not reflect the true cost of achieving the corresponding goals (see Table 4-1). 37 This function combines “micro” levels (purchase from individual providers) and “mezzo” levels (as inter-fiscal transfers to subnational governments). 124 Table 4-1 Budget executed by product unit Population with access With access to Refocused prenatal Normal newborn REGIONS to family planning health counseling care care methods services (SRH) AMAZONAS 14 18 0.4 11 AYACUCHO 43 37 12 93 CAJAMARCA 19 26 3 100 HUANCAVELICA 8 76 5 27 HUÁNUCO 17 34 3 22 PUNO 20 15 5 50 UCAYALI 14 28 0.03 60 NATIONAL AVERAGE 36 42 8 76 Source: WB staff. Note: Budget allocated by product unit: examples of inter-regional variations for some products.  In primary healthcare, the SIS uses the so-called “capitation” mechanism that is based on performance criteria, which would imply criteria related to the demand for services. However, it would seem that the capita is calculated on the basis of the historical production, which is monetized using a combination of tariff plans with historical costs that tend to be determined rather by inputs in the historical supply. In this sense, the concept of the “capitation payment”, which requires a clear definition of the P (Price) and Q (Quantity of people to be treated) is not being applied (see Figure 4-12). Figure 4-12 Transfer of the SIS capitation payment by insured, 2014 90 80 Soles per insured 70 60 50 40 30 20 10 0 Madre de… Huancaveli… Tacna Puno Ancash Callao National Lima Huánuco La Libertad Ucayali Cusco Pasco Moquegua Arequipa Ayacucho Loreto Cajamarca San Martín Junín Ica Tumbes Amazonas Apurímac Source: World Bank.  In the hospitals, the SIS pays per service based on a tariff scheme. In theory, this should lead to a greater production of services by the hospitals. Nonetheless, the following challenges have been identified: (i) The hospitals have acted with discretion to negotiate tariffs with the SIS, and so the tariff scheme is neither uniform nor consistent . In addition to the tariffs approved for the products that are part of the basket of benefits, other expenses incurred by the hospitals are reimbursed as well. 125 (ii) There are no appropriate mechanisms to verify service delivery, in part because of the lack of information systems with which it is possible to cross the billing with other sources of information such as the electronic clinical records, patient records, and others. This lack of auditing opens the door for abuse and overbilling.  Finally, both in primary healthcare and in hospital care, the transfers from SIS to the Execution Units are not included in the budget at the beginning of the year, but at the time of the transfer. Therefore, they cannot be planned duly in advance (see Table 4-2). Table 4-2 Examples of the allocated and executed budget of budgetary programs 2012 2013 2014 Physical target A E A E A E Refocused prenatal care 77 194 129 380 177 226 Normal delivery care 86 576 280 680 282 694 Normal newborn care 76 147 79 174 84 181 Source: Ascarza (2015) based on SIAF. Note: A = Allocated; E = Executed. 28. The RBB, the SIS, and the historical public budget apply different financing guidelines which gives rise to important challenges in execution. The study points to the lack of complementarity among these mechanisms to finance the production factors, i.e. human resources, goods and services, medicines and inputs. Furthermore, the coexistence of mechanisms gives rise to high transaction costs for the service providers, both in primary healthcare and in the hospitals. Finally, it is clear that insufficient attention has been paid to the incentives resulting from these mechanisms for the providers and to the balance between autonomy and accountability. 29. EsSalud is not using modern mechanisms for the purchase of services either and the funds are still being allocated on the basis of inputs. This is directly linked to the deficient separation of the functions of provision and purchase of health services. 30. Finally, the financing mechanisms in the purchase and distribution of pharmaceutical products and inputs give rise to significant inefficiencies. According to MR No. 1753-2002- SA/DM, medicines and medical-surgical inputs are purchased at the national, regional or institutional level, depending on the type of good. In general, so-called strategic medicines and inputs have to be purchased at the national level and are sent to DIRESA, the Specialized Institute or Hospital through the goods remittance modality. The Integrated System for the Supply of Medicine and Medical-Surgical Inputs (SISMED) manages a central revolving fund that centralizes resources from various sources, including resources from the sale of medicines and inputs at decentralized levels, SIS resources and resources of the Program for the Administration of Management Agreements (PAAG). The so-called support medicines and medical-surgical inputs are purchased at the regional or institutional level, using revolving funds at the level of DIRESA or the Execution Unit. These funds are fed by the Program for the Administration of Management Agreements (PAAG), SIS, directly collected resources, and other sources. The government is also responsible for the logistics related to distribution of the medicines and inputs, 126 while the health centers and public hospitals are responsible for distributing these medicines through pharmacies or own counters. 31. The first challenge in this area is the limited competition in the procurement of medicine, equipment, and medical supplies. Without taking into account the procurement processes that are declared deserted, the average number of bidders in the procurement processes is between 1.3 for medicines in EsSalud and 1.9 for equipment and supplies in regional governments. The average number of bidders is one (1) for medicines, equipment and supplies in all subsystems (see Figure 4-13). Observations while preparing this report have shown that the tenders that are unsuccessful at the central level are transferred to the regional level and even to the Execution Units where the transaction costs and prices are higher. In addition, this creates opportunities for collusion among the bidders. Finally, there is considerable variation in the medicine purchasing prices among the various subsystems of the sector (see Figure 4-14). Figure 4-13 Bidders in processes to purchase medicines, equipment and supplies, 2015 Average number of bidders Average number of bidders 2.5 1.2 2 1 0.8 1.5 0.6 1 0.4 0.5 0.2 0 0 Medicines Equipment and supplies Medicines Equipment and supplies Source: World Bank based on SEACE data 2015. Figure 4-14 Price and quantity of antibiotics and anticonvulsants Average price Average quantity 10 150000 Average price by unit 8 Quantity 6 100000 4 50000 2 0 0 Antibiotics Anticonvulsants Antibiotics Anticonvulsants Source: World Bank based on data of SEACE 2015. 32. To date, the government is not using modern mechanisms to buy medicines, e.g. framework agreements. Framework agreements provide for a framework of bidders, prices and 127 competitive conditions for certain products, within which the Execution Units (hospitals, DIRESAS, etc.) can buy without any additional procurement processes. The potential advantages are clear: it is possible to obtain lower prices since the committed quantities are larger compared to the purchases by individual Execution Units; it is not necessary to buy and store medicines and inputs from large procurement operations, since the deliveries would be made directly to the Execution Units where and when needed; and finally, this would reduce the opportunities for collusion and corruption since the price would be more transparent. On the other hand, for certain medicines used to combat HIV/AIDS, certain medicines are purchased in national processes at prices that are higher than the reference prices of the WHO strategic fund (e.g. the price paid for Tenefovir was S/3.35 per unit while the WHO reports a price of S/0.98). The WHO estimates that the additional spending through the national purchase of HIV/AIDS medication was more than S/15 million soles (PAHO 2016). 33. A second challenge is related to the financing and distribution of medicines, giving rise to the misuse of resources. The financing system for medicines and inputs is based on a hierarchy of revolving funds at the central (SISMED), regional (DIRESAs) and institutional levels. Historically, the use of revolving funds was justified to recover part of the costs of delivering medicines in the health centers and hospitals, control the charge to patients and facilitate the processes for different administrative levels to buy medicines and inputs. The system is based on central planning of the needs for medicines and inputs, fed by a hierarchical system of reports on needs from the institutional level to the central level. But this system is not compatible with the concept of health insurance. One of the central goals of the SIS was to eliminate out-of-pocket expenses for the insured, including the purchase of certain medicines. In this sense, the SIS seeks to replace the resources received from patients through the sale of medicines (RDR) by resources coming from the insurance. The SIS became part of the existing system of revolving funds and transfers financing to the regions and the institutions to replenish the medicines used by the people covered under the SIS. In practice, the monitoring system does not allow for linking the provision of medicines to the people insured under the SIS to the SIS transfers. Although the SIS “pays for” the medicine for its insured, this is not an identifiable purchase or transaction – it is an accounting operation within the revolving fund. Since the system does not allow for identifying the person who received the medicine financed with the revolving fund, implicit incentives are being created to sell the medicines (for example to people who are not covered under the insurance or to private pharmacies) and report them as having been delivered to the people insured under the SIS. Expenditure monitoring systems 34. Peru is not yet able to produce basic information about the pillars of the health system. On the one hand, relevant progress has been made in information management in some areas, e.g. the electronic payroll, the National Register of AUS Beneficiaries, and the collection of continuous and reliable household information. Nonetheless, there are major shortcomings in the availability of the information needed for management of the system. For example, there is no information with an acceptable degree of reliability on the number of employees in the health sector or the installed capacity, e.g. the number of hospital beds (Streveler 2014). Most hospitals do not have a modern hospital management system, which gives rise to outdated service management processes, long queues and waiting times for the patients. The HIS design has not been modernized and so parallel information systems have been created, e.g. the device used for 128 SIS insured to get an appointment (FUA). The current situation is that staff prioritize the collection of information in only one system, depending on the patient’s coverage or the incentives. 35. The absence of mechanisms to monitor resource management creates inefficiencies . First, it creates opportunities for the inadequate use of public funds. Second, it hinders monitoring the quality or safety of the patients – for example, there is no reliable information on the prevalence of hospital infections. Third, it hinders the measurement or monitoring of the effective degree of delivery of the PEAS package in the public sector. The duplicity of the systems in a major source of inefficiency that gives rise to information that is neither reliable nor representative. Moreover, this causes malaise among the human resources in the health sector working with the population. In financing, the systems cannot be used for an exact calculation of the expenditure on personnel or primary healthcare. 36. To achieve the interoperability of insurers and providers, it is essential for all stakeholders apply the same information standards and have access to the physical and human infrastructure to exchange information in a fast, safe, and reliable manner . There is a consensus that the current lack of standards and the lack of application thereof is a serious impediment for the country’s capacity to automate its care and insurance processes. In terms of the physical infrastructure, the data exchange networks are slow, the security of confidential health-related data is very inadequate and there is no modern data storage infrastructure for the sector; rather, each institution makes investments in an uncoordinated manner. Finally, the sector lacks an institution that acts as the leader and coordinator. Without these basic conditions to automate the service exchange processes, the interoperability of the insurers and service providers will remain on an artisanal level. Conclusions 37. Improving the efficiency of health spending requires strengthening and modernizing the health management system in three areas: i) improve the prioritization of expenditures; ii) modernize the procurement and payment systems for health care providers; and iii) improve health services planning, execution and monitoring (see Table 4-3). In the medium term, Peru should determine whether the model of the interoperability of insurers and providers can lead to a universal health insurance or whether it is necessary to consolidate and unify the public system with the social security scheme. In the short-term, the challenge is to modernize the system so it would be more efficient, effective and equitable. In terms of efficiency of the spending: 38. Improve prioritization of expenditures: from passive purchases to strategic purchases.  Redefine the package of benefits. In the public system, it would be essential to have clarity about the degree of delivery of the PEAS basic package and the costs of a possible extension of the package prior to broadening it. Moreover, it is necessary to monitor the beneficiaries of the broadened package, ensuring equity in the access. In EsSalud, the scope of the packages of benefits should be rethought in accordance with the collected revenues. The negotiation about the package of benefits of the subsidized insurance and the scheme based on partial contributions should comprise the MEF, the MINSA, the regions and the SIS. This decision should not be a decision taken by the SIS alone. 129  Redefine the system for delivery of the benefits. Peru should develop guidelines on clinical practice that provide for the care paths by type of health condition, with mappings on the levels of primary care, specialized care, complementary services and medium and highly-complex hospitals.  Develop strategic purchasing. The mapping of the care paths should be the basis to develop the strategic purchase of health services administered by an insurance institution, replacing the current system of “passive purchases”. This strategic buyer should (i) participate actively in prioritization of the services to be provided (type and quantity); (ii) define the combination of providers that offer maximum benefits for society; (iii) adjust the provider payment and financing systems to ensure the transition from the current provision system to the desired provision system (adapted from WHO). Some examples of adjustments are 38: o Protect financing of the primary healthcare service and preventive services. o Use incentives to limit the provision of high-cost services. o Pay relatively low prices to the providers for high-cost but low-priority services. o Use groups related to the diagnosis together with an overall budget to control the general spending levels. o Link part of the payment made to providers to their performance. o Introduce co-payments for patients who decide themselves to go to hospital or who skip primary healthcare. o Negotiate the price of medicines, for example by using reference prices or information on the cost-effectiveness ratio.  Enhance transparency and explicitly explain the costs and hidden subsidies among subsystems. EsSalud has not invested in primary healthcare and its insured use the GORE/MINSA services without covering the actual cost thereof. This could be solved by making EsSalud pay for the care at the actual cost. This could be linked also to an equally significant effort of granting the SIS insured access to the high-resolution services of EsSalud, such as the Heart Institute (INCOR) or the Almenara Hospital. These hospitals could function as referral hospitals for the whole Peruvian population. 39. Modernize the procurement and payment systems for health care providers.  It is necessary to revise the system of financing and payment to the providers in order to align the incentives to the expected health outcomes. In the long term, a system of strategic purchases by a public insurer should provide appropriate financial incentives for the providers to be aligned to the desired organization of the system, i.e. assigning responsibility. 38 Source: WHO. 130  The Register of Beneficiaries of the Universal Health Insurance could be the basis of a financing strategy thereof. For this to be possible, a previous phase should standardize the insurance concept and align the foundations of the different insurance institutions.  In the short term, it would be necessary to solve the existing duplication between the RBB and the SIS payment systems. The existing arrangement through which the SIS provides partial financing while the other public funds finance most of the service cost, does not provide clear incentives to the providers. The payment systems used by the SIS could be modernized and be made more transparent. Even though the RBB has been effective in ensuring the availability of financing for the prioritized lines (such as maternal and child health or nutrition), this would not lead to an efficient resource planning. In order to enhance efficiency, the current micro management of the expenditure should move forward to payment mechanisms that incentivize the production of health services and outcomes.  It should be noted that there is considerable international experience in the field of payment systems which Peru should take advantage of. It is important to consider the balance between autonomy and accountability in the payment systems.  The current system for the purchase and distribution of pharmaceutical inputs and products is not very efficient or effective. In the medium term, Peru could begin with the transition from a system in which the Government buys and distributes medicines to a more modern third party buyer reimbursement system. In the short term, without changing the system, the efficiency of the purchase and distribution of medicines and inputs could be improved: i. Defining the technical characteristics of the medical inputs and equipment that could be homogeneous in all hospitals and public centers in order to enhance cost-efficiency of the purchases; ii. For the medicines, inputs and equipment that account for a large part of the expenditure or the purchase price of which display major variations, the option of buying under framework agreements could be considered; iii. Modernizing the systems for delivery and monitoring of the distribution of medicines, using the patient’s identity document (DNI) for identification. 40. Improve health services planning, execution and monitoring.  To date, the exchange of services between EsSalud and the public system is virtually inexistent in practice. For the exchange to take off, it will be essential to enhance transparency of the existing subsidy between the public system and EsSalud, harmonize the used service procurement systems, generate incentives for the primary care units of the public sector being able to bill EsSalud and solve the pending issue of the debt of regional and municipal governments with EsSalud.  In order to move forward towards a true interoperability of the insured and the providers, the used systems and procedures should be compatible and/or standardized. Without standardized costing or service classification methodologies, the exchange of services would remain on an 131 artisanal level and not have the systematic impacts on the services which the population expects.  Peru would need to build a modern health information system as the basis for the universal insurance which would bring benefits in terms of access, equity, efficiency and quality. In particular, Peru should invest in: (i) a dictionary of health data, i.e. a comprehensive list of standards to be used in the classification of patients, providers, services, etc., which is used in all subsectors, including the public sector, social security and the private sector; (ii) a safe health communications network which providers, insurers and verifiers can use to safely transmit and exchange confidential health information; (iii) modern data warehousing centers equipped with the necessary software to analyze and interpret the data; (iv) a national health data center that is responsible for proper functioning of the health information systems. Policies options Below is an overview of policy options that may lead to gains in efficiency, equity, and potential fiscal savings (see Table 4-3). 132 Table 4-3 Policy options to improve efficiency and equity and generate savings Policy context Proposal Efficiency Equity Potential Term saving Provision and insurance EsSalud and the health funds of the armed Consolidate the schemes of + Neutral 0.1-0.2% of Long term forces and the police provide parallel EsSalud and the ‘sanidades’ GDP insurance schemes and services within the and/or establish mechanisms framework of social security. This gives rise for exchanging services. to duplicity of the infrastructure and administration. The EsSalud insured use public health Consolidate the register of the Neutral + 0.1% of GDP 1-2 years services without the public system billing the universal health insurance; use social security scheme for these services. it to register the services (Estimates are 5 percent of the primary provided to the EsSalud healthcare users and 10 percent in hospital insured, starting with hospital care) care. (MINSA) There are no clear guidelines for Develop clinical practice guides + + Neutral 1-2 years clinical practice which enhance efficient with care paths, by type of health services for patients among different care condition. levels and types of providers. (MINSA) The mechanisms for the Develop the strategic purchase + + Neutral 2-3 years financing/payment of providers do not of health services, gradually comprise “strategic purchases” to incentivize incorporating the different an efficient service delivery – on the contrary, sources of financing. the hospitals are incentivized to provide Move to a training system with + + Neutral primary level care. The different financing performance criteria in primary modalities (historical budget, SIS, RBB) healthcare. generate cross incentives for the providers. 133 Policy context Proposal Efficiency Equity Potential Term saving Pharmaceutical products and inputs For many corporate purchase items at the Use the mechanism of + Neutral + 1 year national level, few providers submit bids, i.e. purchases based on framework there is little price competition. The tenders agreements to avoid that collapse at the national level are overcharges in purchases at the transferred to the regional level and even to level of the Execution Units. the level of the Execution Units where the transactions costs and prices are higher. Moreover, this creates opportunities for collusion among the bidders. The electronic reverse auction mechanism is Replace the presence-based + Neutral + 6 months not used for the purchase of pharmaceutical reverse auction mechanism by products and inputs at the national level. the electronic reverse auction mechanism. The option of purchases through the WHO Check which tenders could ¿? Neutral + 6 months Strategic Fund (pooled purchases at the take place through the Strategic international level that ensure better prices) Fund. are not taken advantage of. The delivery channel of pharmaceutical (Short term) Improve the + + + 1-2 years products in the public sector is inefficient. The medicine distribution system to combination of revolving funds and the limit inappropriate use and distribution of fixed quantities does not allow waste (e.g. by linking the for the provision of pharmaceutical products distribution of medicines to the to be in line with the needs of the health supply level, improve the centers. This gives rise to shortage and information on distribution) – simultaneous supply, the wasting of expired electronic and transparent medicines and misappropriation. stock management. (Medium term) Invest in a computer system that allows for moving forward towards a 2-3 years distribution system based on 134 Policy context Proposal Efficiency Equity Potential Term saving pharmacies and third party buyer reimbursement (modern health insurance system). Hospitals Public hospitals do not have a modern hospital Procure a hospital information + + + 1 year administration system. They do not have system to (i) rationalize the use reliable data on efficiency in the use of beds, of times of human resources, the duration of hospital stays, the use of inputs and hospital efficient procedures, waiting times, intra- infrastructure, and (ii) enable hospital arrangements, etc. shift assignment and decrease queuing and waiting times for patients. The care model is not very efficient and does It is necessary to promote the + Neutral No specific Medium term. not take advantage of the existence of use of more efficient and data are technologies to lower unit costs (e.g. higher-quality care models, available for outpatient laparoscopic surgery, high-volume e.g. day hospital, outpatient Peru. laboratories). surgery, laparoscopic surgery, International the centralization of reference: laboratories by cities and/or Decrease of regions, the centralization of costs by 18- medical imaging services, etc. 68 percent by moving from a selection of inpatient surgeries to outpatient surgeries. The deficient exchange of medical Invest in computer systems to + + + information among providers / care levels in improve the availability of EsSalud gives rise to duplicity and the overuse medical information. of medical examinations. 135 Policy context Proposal Efficiency Equity Potential Term saving Package of benefits The package of benefits of the EsSalud Prepare a package of benefits + + + regular insurance does not have exclusions. based on the cost-effectiveness of pharmaceutical products and procedures. The SIS package of benefits has been Provide for financing of the + + Neutral <1 year broadened in theory but is not in line with the package components, prioritize delivery capacity and does not have services according to their transparent financing. effectiveness and the disease burden. 136 Introduction 1. Peru has made significant progress in the education sector, though it is still lagging behind comparators. Between 2000 and 2015, Peru’s results within the framework of the Program for International Student Assessment (PISA) showed the greatest improvement of any country. Despite this positive step, education in Peru does not only lag behind the OECD countries but also in terms of what would be expected of the country according to its income level. Despite the progress in net enrolment rates, only one out of every four 15-year-old has math skills to take part effectively in productive activities, compared to an OECD average of 77 percent. Moreover, only one out of every three secondary school students reach the PISA basic level in reading comprehension (compared to 83 percent in the OECD countries). Additionally, the education in socioeconomically poor schools is far from the learning in environments with more resources. Peru has one of the deepest learning gaps by socioeconomic level among the PISA participants. The country has begun to carry out several reforms in the education sector to improve the learning outcomes. 2. Even though Peru invests less than the OECD average in education it could use the available budget more efficiently. The school infrastructure requires interventions, enlargement, better ICT and laboratory equipment, and maintenance. At the same time, the sector has important inefficiencies, especially in some regions. It is important to correct these inefficiencies and improve the capital and current spending processes within the context of growing spending pressures in public education. 3. This chapter evaluates the efficiency and effectiveness of public spending in education in Peru. Potential efficiency gains are identified in the expenditure through a frontier analysis, paying special attention to the differences among subnational governments. The efficiency analysis is based on global evidence about what works well in education. The sustainability analysis provides a crosscutting vision of the challenges concerning equity. The chapter is organized in three sections. The first section provides a panorama of the expenditure on education, as well as the distribution thereof and recent trends. The second section presents an analysis of the efficiency and effectiveness of the expenditure on education. The last section presents recommendations to improve the expenditure on education to maximize its impact and reduce inequalities. Expenditure and Results Expenditure 4. The expenditure on education has grown since 2011, just like the rate of spending execution. After a declining trend between 2009 and 2011, the expenditure increased from 3.4 percent of GDP in 2011 to 4.1 percent in 2015, and the execution of this expenditure increased from 82.3 percent to 90 percent (see Figure 5-1). In this period, the expenditure on education as a percentage of total public spending increased from 14.9 percent to 17.1 percent. In absolute terms, it increased from S/13,708 million in 2011 to S/22,796 million in 2015. Figure 5-1 Public budget for education and rate of execution, 2011-2015 92.0 4.09 4.50 3.93 3.83 90.0 3.51 3.58 3.64 4.00 3.40 Rate of execution (% of PIM) 90.0 3.50 88.0 88.6 3.00 Percentage of GDP 86.0 2.50 84.0 85.7 85.1 84.8 2.00 84.0 82.0 1.50 80.0 81.2 1.00 % PIM 78.0 0.50 % PBI 76.0 0.00 2009 2010 2011 2012 2013 2014 2015 Source: SIAF and BCR. Note: The public budget is calculated on the basis of the PIM, as a percentage of GDP. The rate of execution is calculated based on the expenditure executed as a percentage of the PIM. 5. Expenditure as a percentage of GDP have increased for all levels of basic education and, to a lower extent, also for higher education. Between 2011 and 2015, the public expenditure on primary and secondary education grew by around 35 percent in real terms. In both cases, the increase was largely in response to the growing enrollment. In higher education, there was an increase as well (see Figure 5-2). A greater proportion of the expenditure in infrastructure has been allocated to the different levels of education, explaining the increase in expenditure at different levels and the reduction of the direct allocation to basic infrastructure. Figure 5-2 Public spending as a percentage of GDP, by type of service, 2009-2015 Source: SIAF. Own elaboration. Notes: (1) SBE and ABE = Special Basic Education and Alternative Basic Education. (2) Higher and vocational education include public and private universities, as well as vocational training and educational programs to promote employment. (3) The expenditure on basic infrastructure refers to the expenditure on infrastructure in Basic Education which is not allocated to a specific level. 6. Despite this progress, Peru spends less on education than its peers. However, the share of education expenditures in total public expenditures has increased and is relatively high compared to other countries due to Peru’s overall relatively low public expenditure levels. The country spends less than the OECD countries and the regional and structural peers in absolute 138 terms (dollar PPP) and as a percentage of GDP (see Figure 5-3). Peru spends PPP$ 1,495 per student in primary school, PPP$ 1,562 in secondary school and PPP$ 972 in higher education, which is one fourth, one fifth and one tenth of the expenditure by other OECD countries, respectively. However, it is interesting to note that the share of education in total public spending is larger in Peru than in the other OECD countries. Seemingly, the limitation of the expenditure on education is related more to the overall level of public spending than to its relative size. Figure 5-3 Public expenditure by student, by level of education, 2012-2014 In PPP$ As a percentage of GDP per capita Source: WDI: UNESCO Institute for Statistics. Structural peers (2013): Canada, Australia, Malaysia, Poland, Rumania, Thailand. Regional peers (2013): Brazil, Colombia, Mexico and Chile. Data of Peru (2014) and OECD data (2012). 7. The way in which the country distributes its education budget is different than in peer countries. Almost 40 percent of the resources are allocated to primary education. This is much higher than the other OECD countries. On the other hand, Peru allocates one sixth of its education budget to the tertiary level, while some structurally comparable countries and OECD members allocate one fourth of their budget. This may be explained by differences in enrollment rates on the different educational levels. In terms of volumes, the spending on education in Peru is low on all levels relative to its peers (see Figure 5-4). 139 Figure 5-4 Comparison based on distribution of the expenditure by level of education, 2012-2014 Source: WDI: UNESCO Institute for Statistics. Structural peers (2013): Canada, Australia, Malaysia, Poland, Rumania, Thailand. Regional peers (2013): Brazil, Colombia, Mexico and Chile. Data of Peru (2014) and OECD data (2012). 8. The main category of spending in education is wages. According to international data, between 2012 and 2014 around half of the (direct) expenditure in educational institutions on all the levels was for wages (the highest level was in secondary education, with 60 percent). In contrast, its structural, regional and OECD peers allocate between 73 percent and 82 percent to wages in basic education and between 54 percent and 61 percent in higher education (see Figure 5-5). Nonetheless, this amount does not include administrative services contracts (CAS), a common modality to hire staff outside the payroll which accounted for 4 percent of the total expenditure on education in Peru in 2015. Neither does it include the payments made to facilitators and literacy educators, who work in out-of-school preschool education (PRONOEI) and alternative basic education (ABE). 140 Figure 5-5 Expenditure on wages as a % of the total spending in public schools (IIEEs), 2012-2014 Source: WDI: UNESCO Institute for Statistics. Structural peers (2013): Canada, Australia, Malaysia, Poland, Rumania, Thailand. Regional peers (2013): Brazil, Colombia, Mexico and Chile. Data of Peru (2014) and OECD data (2012). 9. The expenditure on personnel as a share of GDP in 2015 was lower than the expenditure share in 2009, even though the share has increased since 2011. For this analysis, we also consider the expenditure on CAS and the payments made to facilitators and literacy educators. In 2015 the total expenditure on personnel was 0.13 points less of GDP relative to 2009, even though the expenditure on CAS increased by 0.1. Naturally, the main personnel category was the educators (teachers, facilitators and literacy educators), which accounted for between 70 percent and 74 percent. The expenditure on administrative personnel on the payroll and CAS personnel accounted for 16.9 percent of the expenditure on personnel in 2015, compared to 12.6 percent in 2011, due to the increase in CAS. The remaining 10 percent was invested in pensions, benefits and other social obligations (see Figure 5-6), mostly for funerals, because in that year the government paid an important social debt in this item. 141 Figure 5-6 Expenditure on personnel, obligations as a % of GDP, by educational level, 2009-2015 2.00 1.50 1.00 0.50 0.00 2009 2010 2011 2012 2013 2014 2015 Teachers & Assistants Administration ACS PBOS Source: SIAF and BCR. Own elaboration. PPOS: Pensions, benefits and other social obligations. 10. In recent years, an effort has been made regarding investments in development, procurement, and the delivery of materials to schools. One particularly outstanding aspect is the increasing expenditure on educational material for bilingual preschool and primary education, which increased by S/128.48 million between 2012 and 2015, compared to S/2.41 million between 2007 and 2011. Between 2012 and 2014, the secondary school level received an important amount of materials, with the greatest increase taking place in 2013 (see Figure 5-7). As from 2014, the expenditure in the three levels continued to be relatively high, compared to the expenditure in 2011 (around three times higher) and the special basic education sector received materials as well. Simultaneously, there were improvements in the level of budget execution of this expenditure item, one of the most complex ones in terms of procurement processes, from 84 percent (expenditure relative to the PIM) in 2011 to 99 percent in 2015. The greater expenditure in materials was translated into more timely deliveries, with an increase from 13.2 percent to 65.1 percent between 2011 and 2014 as regards the percentage of schools receiving materials on time (INEI, 2011-2014). This seems to be the primary factor explaining the improved learning achievements, as is shown below (Castro, 2016). 142 Figure 5-7 Current expenditure on educational material in Basic Education as a percentage of GDP, by type of educational material and modality, 2011-2015 By language of instruction By educational level Provision of IBE material Provision of L1 material in Spanish Pre-primary Primary Secondary Special 0.12 0.08 Expenditure executed as% of GDP 0.10 0.07 0.06 0.08 0.05 0.06 0.04 0.04 0.03 0.02 0.02 0.01 0.00 0.00 2011 2012 2013 2014 2015 2011 2012 2013 2014 2015 Source: Kudo & Székely, 2015 (prepared on the basis of SIAF). Note: Updated with the budget execution as at December 2015 and with the executed expenditure (accrual basis). 11. There is a significant upward trend in access to basic education (especially preschool education) and to higher education based on the Beca 18 program. 39 In 2012, the budget for actions to enhance the coverage of basic education (regular, special and alternative) totaled S/19 million (0.0038 percent of GDP) and by 2015 this amount increased to S/694 million (0.11 percent of GDP). The main expenditure item explaining this increase is the upgrade and enlargement of the infrastructure for preschool education, which increased by 24 times, from S/17.9 million to S/424.4 million. In higher and vocational education, the expenditure on the access to and provision of services (beyond the basic inputs) increased both relative to GDP (by 0.1 pp) and relative to the total expenditure on education (by 1.8 pp). This is primarily thanks to introduction of the Beca 18 program (S/593 million). 12. The proportion of the capital expenditure relative to the total education expenditures is higher in Peru than in other countries at all education levels. In preschool education, the capital expenditure in Peru accounts for almost 24 percent of the total expenditure in schools, compared to 11.5 percent in OECD countries and in countries of the region (see Figure 5-8). The situation is similar in primary and secondary education, while the capital expenditure in higher education in Peru is similar to its structural and regional peers but twice as high as the OECD average. 39 The Beca 18 program targets youth who completed secondary school (public and private schools) and who want to go to university. 143 Figure 5-8 Capital expenditure as a % of the total expenditure in public schools, by educational level Source: WDI: UNESCO Institute for Statistics. Notes: Structural peers (2013): Canada, Australia, Malaysia, Poland, Rumania, Thailand. Regional peers (2013): Brazil, Colombia, Mexico and Chile. Data of Peru (2014) and OECD data (2012). 13. At first sight the high share of capital spending in the total spending is a welcoming development but may also reflect inefficiencies. A substantive capital allocation in the sector points to the importance given to improvement of the physical infrastructure and the educational equipment for the students. This is very positive and part of the trend in the country. On the other hand, it also reflects inefficiencies in terms of maintenance of the existing infrastructure and it exercises pressure in terms of an improved prioritization of new investments. The current infrastructure suffers from a greater deterioration due to a lack of maintenance and/or renovation (World Bank, 2016) compared to the OECD countries, structural peers and even some regional peers. The absence of a medium-term strategic plan that prioritizes investments based on equity and cost-effectiveness criteria to optimize investments does not help. Furthermore, the unsatisfied demand in terms of infrastructure is greater in Peru40 than in comparator countries. 14. Investments in pedagogic quality would improve the effectiveness of basic inputs spending. This expenditure has virtually tripled relative to GDP and has doubled as a percentage of the total expenditure in education, though it continues to be relatively low: 4 percent of the expenditure in education in 2015, and it continues to be quite centralized. The growth is related to the initial actions for implementing the Public Teaching Career (CPM) and creation of the program for Pedagogic Support in primary school, primarily focusing on integral urban and poly-teacher schools. Ninety percent of the expenditure of the Education Units (UEs) of the Regional Educational Directorates (DREs) and of the Local Educational Management Units (UGELs) 41 is for wages and 5 percent for materials and infrastructure (see Figure 5-9). This means that less than 5 percent is left for interventions to strengthen the pedagogic quality. 42 40 Estimates are that the investment required for the educational infrastructure to have capacity to satisfy the unsatisfied demand is almost S/23 billion, without including university education (World Bank, 2016). 41 The DREs and UGELs are part of the Ministry of Education and are responsible for coordinating, directing and evaluating the educational and administrative activities of public and private institutions in Peru. 42 This includes curriculum management and the definition of learning progress maps and standards, in-service training activities, training and specialization of the school personnel, coaching and pedagogic support, the design of 144 Figure 5-9 Regional average expenditure by student in UEs of DREs and UGELs, 2015 Source: SIAF and School Census. Executed current and capital expenditure, excluding the expenditure to increase coverage. Does not include indirect expenditures (planning and management). Results-based management is related to management of the PELA program. 15. The contribution of the different government levels to total spending varies significantly. Regarding distribution of the expenditure in the territory, Lima Metropolitana concentrates 13.5 percent of the total expenditure and a similar percentage corresponds to the expenditure at the national level. The national expenditure is related to policies and interventions carried out in the whole country and is executed entirely by the MINEDU, SUNEDU and SINEACE. In Lima Metropolitana, almost one third of the public expenditure is executed by the universities and the rest by the MINEDU through the DREs and UGELs. In the other regions, the expenditure comes mainly from the regional government but with major variations among the regions: from 78 percent in Loreto to 55 percent in Huánuco (see Figure 5-10). The MINEDU participation is greater in some regions than in others: from 133 million in Amazonas (25 percent of the region’s spending) and 121 million in Cajamarca (10 percent of the region’s spending), to 2 million in Moquegua (1.2 percent). On the other hand, the local governments have a participation that varies from 219 million in Cusco to 4 million in Madre de Dios. interventions and models for schools with particular needs, as well as all actions related to implementation of the CPM in schools and the processes and projects related to quality assurance in higher education. 145 Figure 5-10 Participation of government levels in the public expenditure on education, by region 3,500 MINEDU* University GR GL 3,000 2,500 2,000 1,500 1,000 500 Huánuco Amazonas Lambayeque Madre de Dios Ucayali Ayacucho Puno Tumbes Cajamarca Piura Ica Tacna Callao Cusco Apurímac Lima Metropolitana Moquegua Nacional Ancash Junín La Libertad Lima Provincias Loreto Pasco Arequipa San Martín Huancavelica Source: SIAF. Own elaboration. Notes: The national expenditure is the one made by the central government for the benefit of the entire country or various regions. MINEDU includes the expenditures of SUNEDU and SINEACE. 16. The allocation of educational resources does not take into account that the costs and challenges regarding education are greater in more remote areas. As shown in Figure 5-11, there is a negative relationship between the average distance between the schools and the UGEL (in minutes) and learning, i.e. the farther away the schools, the higher the number of students with lags in learning. Nonetheless, the average expenditure by student by UE is not related to the level of dispersion of the schools covered by the UEs (see Figure 5-11). 17. Expenditures per student vary substantially among regions. For example, the average expenditure on primary education by student is three times higher in Moquegua and Huánuco than in Loreto and Ucayali. In some regions, the expenditure by student is much lower than the level required to deliver quality services. For example, in Loreto and Ucayali, two forest regions, the provision of services and the retention of qualified teachers are more expensive, particularly if the idea is to effectively cover the need for bilingual teachers trained for BIE. 146 Figure 5-11 Expenditure by student in primary school and learning achieved in mathematics, by UE, 2015 Source: SIAF, ECE 2015 (2nd grade of primary school) and School Census 2015. Own elaboration. Note: For a better visualization, the figures exclude the UEs Dátem del Marañón and Nauta because the average times in their schools are extreme (1,171 and 1,686 minutes) even though they follow the general trend; they are counted for the polynomial adjustment (R2). Outcomes 18. The learning outcomes in primary and secondary schooling have improved significantly since 2009. According to the national test for students in the second grade (ECE) developed by the Ministry of Education (MINEDU), which is part of the universal testing strategy, the proportion of students with a satisfactory reading level increased from a very low 15.9 percent in 2007 to 50 percent in 2015. In mathematics, the percentage with a satisfactory level has tripled, from 7.2 percent to 26.6 percent of the students (see Figure 5-13). Furthermore, the performance among secondary school students has improved as well. Between 2000 and 2015, Peru’s outcomes within the framework of the Program for International Student Assessment (PISA) showed the greatest improvement of any country. 19. But there is still considerable margin for improvement: almost half of the secondary school students do not know how to read or do not understand what they read, and almost three quarters do not have a satisfactory level in math (level 1 or lower in Figures 5-12, 5-13). Despite the important improvement in the PISA outcomes in 2015, the outcomes in Peru are still very low and display the worst level in South America. Only six countries (Lebanon, Tunisia, Macedonia, Kosovo, Algeria and Dominican Republic) out of the 73 countries that participated in the program had lower results. 20. The access to higher education has improved, but not for all. The graduation from higher education among youth (22 to 24 years old) has increased from 12 to 20 percent between 2003 and 2013. While 20.2 percent of the non-poor in this age group have completed higher education, only 8.1 percent of the poor youth and 3.3 percent of the extremely poor have completed higher education. 147 Figure 5-12 Percentage of primary school Figure 5-13 Performance of secondary students with a satisfactory level in ECE, school students in reading and 2007-15 mathematics, PISA, 2015 60 Peru 6.4 47.5 49.8 50 Reading Chile 27.1 Students with HEADCOUNT (%) 40 level 1 or less 26.6 OECD 18.8 30 Students with 20 15.9 Peru 37.7 28.4 level 2 or more Mathematics 10 7.2 Chile 23.0 26.3 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 OECD 14.9 Reading Mathematics 100 0 100 % Source: WB with OECD data. Source: WB with OECD data. 21. There are large differences in learning outcomes among different regions in Peru, reducing the average quality of education. Peru has one of the lowest PISA scores in English and mathematics, largely due to the differences in the quality of education between the provinces and the school districts. In effect, Peru has one of the largest gaps in the learning outcomes among the countries participating in the PISA assessment. In order to improve the average quality of education, Law 29944 (Teachers’ Reform) of 2012 reformed the existing law of teachers to promote implementation of a merits-based system to hire and promote teachers throughout their career. This law encouraged and rewarded high performance and sought to attract more qualified candidates aimed at improving the teaching quality. However, it did not directly address the difficulties to attract qualified teachers and principals to the schools in the poorest areas of the country, particularly outside Lima and Callao. For example, between 70 and 90 percent of the teachers in the rural areas were hired without being included in the teaching career system. The schools in disadvantaged zones also lack a qualified school management. Efficiency and Effectiveness of the Expenditure on Education Efficiency of the public budget 22. Education spending is relatively decentralized and budget execution rates are still relatively low at the local level and in universities, despite recent improvements. In 2015, the central government covered 25.4 percent of the spending in education (versus 21.4 percent in 2011), without including the universities that managed 17 percent. The regional governments concentrated 46.4 percent of the expenditure (versus 47.6 percent in 2011) and the local governments around 11 percent (see Figure 5-14). The rate of execution is higher in the regional governments and very low in the local governments (see Figure 5-14). In part, this may be explained by the composition of the expenditure at each level. While the spending in education of the local government is mainly concentrated (98 percent) in investment projects that have greater obstacles for execution, the regional governments’ expenditure on education is mostly concentrated in wages (78 percent) with regular payments. The bulk of the expenditure on goods 148 and services, i.e. inputs and activities to leverage quality improvements, is executed by the central government and the universities. It should be noted that the public universities have an aggregated level of execution that is lower than that of the regional government, which requires a more in- depth analysis than is possible in this study. Figure 5-14 Public budget in Education and rate of execution by level of government 4.50 100% 98% 4.00 96% 0.70 3.50 0.71 0.75 90% 0.63 0.71 0.76 0.46 3.00 0.48 0.69 0.43 80% 0.39 0.47 0.50 2.50 0.37 74% 2.00 1.90 70% 73% 1.92 1.69 1.80 1.50 1.62 1.67 1.73 60% 1.00 0.50 0.81 0.80 0.74 0.86 1.04 50% 0.73 0.66 0.00 2009 2010 2011 2012 2013 2014 2015 2009 2010 2011 2012 2013 2014 2015 Central Regional Central Regional Local Universities Local Universities Source: SIAF. Own elaboration. Efficiency of expenditures 23. Because of its low spending, Peru is near the efficiency frontier in terms of primary education compared to other countries. With an average expenditure on education similar to that of Panama, Paraguay and the Dominican Republic, Peru obtains better outcomes in mathematics in the TERCE tests43 of 2013 (see Figure 5-15 for the third and sixth grades of primary school). In effect, using the sample of the 13 Latin American countries that participated in the TERCE 2013 and using the methodology of the Data Envelopment Analysis (DEA), we have calculated the efficiency frontier for the learning in mathematics and language for the third and sixth grades of primary school, as well as for sciences in the sixth grade of primary school. 43 TERCE is a Latin American assessment of students in the third and sixth grades of primary school that measures their aptitudes in reading comprehension, mathematics and natural sciences. It is not applied regularly. 149 Figure 5-15 Efficiency frontiers of the public expenditure in primary education in mathematics, TERCE Third grade of primary school Sixth grade of primary school Source: TERCE 2013. Own elaboration. Similar figures have been prepared for reading comprehension and sciences, with similar results. Figure 5-16 Efficiency frontiers of public spending in secondary education in mathematics, PISA Source: PISA 2015 and expenditure by student in secondary education, UNESCO. Similar figures were prepared for reading comprehension and sciences, with similar results. 150 24. The efficiency frontiers method mechanically shows a relatively efficient level of spending per student in secondary education since Peru’s spending levels are among the lowest of all countries included in the analysis. Peru spends less per student on secondary education than most other countries. Some other countries like Albania spend less on education per student but obtain better results (see Figure 5-17). From a global perspective, there is a potential space for efficiency gains of up to 20 percent in mathematics, 11 percent in reading and 22 percent in sciences, with which Peru would continue to be among the countries with a low performance in the test. If it increased the investment in education to 6 percent of GDP, as provided for in the National Agreement, according to this methodology, it would be possible to achieve additional improvements in the students’ learning of around 6-9 percent in mathematics, reading comprehension and sciences. But this would happen only if the additional resources are used efficiently. 25. Strikingly, regions in Peru that spend more on primary education do not necessarily have better learning outcomes. Expenditures and the outcomes in primary education among regions in Peru are not correlated (see Figure 5-17). In many regions, better results have been achieved in 2015 compared to four years before for every sol invested. The regions at the frontier have changed: Moquegua which delimited the upper end in 2011 was replaced by Tacna in 2015. 44 At the same time, there is no positive correlation between the expenditure and outcomes at the inter-regional level. Some regions spend more than twice per student on primary education without obtaining better results, which suggests that there are major inefficiencies at the regional level. 44 For preschool education, the efficiency was calculated against net coverage of the level (population of 3-5 years old enrolled in preschool education). For higher education against timely completion (population of 22-24 years old that has completed higher education). The efficiency frontiers for the other educational levels show qualitatively similar patterns, but they are not reported because of space constraints. 151 Figure 5-17 Efficiency frontiers of the regional expenditure on primary education, comparing 2011 and 2015 60 2011 TAC 2015 50 % of students with a satisfactory performance in mathematics MOQ 40 CALL ICA AMZ PIU ANC PUN 30 JUN PAS AYC TAC LIM MOQ CUS LPR CAJ LIB LAM APU HCO 20 ANC TUM SMA ICA LPR MDD AQP LIB CALL JUN HNV PIU LIM LAM AMZ CUS APU 10 CAJ TUM UCA MDD SMA HCO PUNPAS HNV UCA AYC AQP LOR LOR 0 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Average expenditure on primary education, by student Source: ECE 2011, ECE 2015, SIAF and ESCALE. Note: The expenditure for each level includes the expenditure on infrastructure of a specific level that is not accounted for under the function of the said level. This allows for a more real analysis of the expenditure by student. The expenditure in Lima Metropolitana corresponds to the UEs of the DRE and UGEL of Lima. 26. After controlling for region-specific factors driving the costs of proving education, two regions stand out in terms of their low spending efficiency: Madre de Dios and Tumbes. We have calculated the “net efficiency” by controlling for the incidence of poverty in the region and the aggregated distance of the schools to the UGEL in minutes, since both factors have an impact on the challenges, and therefore on the cost of educating in every region. Map 5-1 shows the regional net efficiency with respect to the learning outcomes in primary and secondary school.45 45 The analysis was made also for preschool and higher education in relation to the coverage of these services. 152 Map 5-1 Net efficiency of public spending in education with respect to learning Primary Secondary Source: WB. Note: The net efficiency was calculated by controlling the incidence of poverty and the distance of the schools to the UGEL in minutes (adding up the time to the schools of every level). The efficiency of primary education was calculated on the basis of current public spending in primary education, by student (2014-2015) and the percentage of students with satisfactory results in mathematics (total percentage) per region. The efficiency in secondary education was calculated on the basis of current public spending in secondary education, by student (2014-2015) and the percentage of students with satisfactory results in mathematics (total percentage) per region. 27. If all regions performed at the national efficiency frontier for given levels of current expenditures, learning outcomes in preschool and primary education would improve by more than 10 percent (see Figure 5-18). The calculation is that the improvements would be smaller in secondary education, in part because the frontier is relative and the results of the first evaluation in 2015 were homogeneously low among the regions. On the other hand, the improvements would be more notable in preschool education than in higher education (see Figure 5-18) because of the same reasons. 153 Figure 5-18 Learning achievements if the regions performed at the net efficiency frontier Source: WB. Note: The efficiency of preschool education was calculated on the basis of total public spending in education, by student (2012-2013) and the net enrollment rate in the level for the age group of 3-5 year olds (total percentage) per region. The efficiency in higher education was calculated on the basis of the average public spending in education, by student (2000-2014) and the rate of completion of higher education for the age group of 22-24 year olds (percentage of the total), per region. Data of 2015 are used for primary and secondary education and data of 2014 for preschool and higher education. 28. The analysis at the level of the UEs shows three important results regarding the relationship between efficiency, dispersions, and lags in learning outcomes (see Figure 5-19). First, there is an exponentially negative relationship between the level of efficiency in a UE and the percentage of students who begin their learning in mathematics (below level 1). In other words, the efficiency does not only result in more students achieving the expected learning, but in fewer 154 students lagging behind. This is good news. Second, with the same level of efficiency, the results tend to be lower as dispersion in a jurisdiction increases. Therefore, the big bubbles in the figure are located higher up than the small ones. Nauta illustrates this: the UE with the greatest geographical dispersion in primary education is number 29 out of 182 UEs in net efficiency, but in terms of the percentage of students who begin, it is number 175. This suggests that it would be necessary to allocate resources that respond to the actual costs of providing education in the poorest, most disperse, bilingual zones or in contexts of violence. Figure 5-19 Efficiency and lag in learning by UE, based on geographical dispersion, ECE 2015 90% Mariscal 80% Ramon Datem Del Castilla Marañon Requena 70% Alto Nauta Amazonas Contamana 60% Loreto % of students who begin 50% Huarmaca Pataz Sanchez Carrion 40% Oxapampa La Convencion Celendin 30% Jaen Amazonas San Ignacio 20% Cusco Morropon 10% R² = 0.6156 0% -0.600 -0.400 -0.200 0.000 0.200 0.400 0.600 0.800 Net efficiency of the expenditure on primary education, 2015 Source: WB. Note: The UEs that are labeled and colored are the ones with the greatest distance to cover, as shown by the size of the circles. The distance to be covered is calculated as the sum of the minutes between the UGEL and the primary school: the more schools and the greater the dispersion, the greater the distance to be covered. They are located mainly in Loreto, Cusco, Cajamarca, Piura, Amazonas and Pasco. Effectiveness of the expenditure 29. What works in education? Which factors explain improvements in learning? To answer the first question, we will show the results of an exhaustive review of literature and recent meta analyses that explain the possible impacts of various interventions. We will do so in a comparative manner, based on the impact of similar interventions in the world 46. In the second 46 See for example: Identifying Effective Education Interventions in Sub-Saharan Africa: A Meta-Analysis of Rigorous Impact Evaluations, Conn (2014); School Resources and Educational Outcomes in Developing Countries: A Review of the Literature from 1990-2010, Glewwe et al. (2014); The Challenge of Education and Learning in the Developing World, Kremer et al. (2013); Quality Education for All Children? What Works in Education in Developing Countries, Krishnaratne et al. (2013); Improving Learning in Primary Schools of Developing Countries: A Meta- 155 part, we will show the results of an analysis of decompositions, exploring the factors associated to learning improvements among students in the second grade of primary school (see Table 5-1). Table 5-1 Possible impacts of interventions with incentives and rewards for a good performance Country, Evaluation Finding Comments Period Selection/Schools with a High Performance Kenia Lucas & Positive effects of 0.27 SD in test scores, but not significant, of In high-performing secondary schools Mbiti (2014) going to an elite secondary school in Kenia. Rumania Pop-Eleches Effects in terms of passing the secondary school leaving exam In high-performing secondary schools & Urquiola between 0.02 SD and 0.1 SD. In addition, it is mentioned that the (2013) stratification of the schools based on quality results in the parents reducing their efforts and the teachers being ordered with the preference of students with a high performance Kenia Duflo et al. Effects on students of the upper half of the performance Organization of classes based on (2011) distribution who received training of 0.19 SD in test scores, while performance in the lower half this was 0.16 SD. These results are persistent. On the other hand, in schools that did not receive training, only the students of the upper part of the distribution improved their scores if assigned randomly with high-performing students School Bonus Gambia Blimbo & Effect of -0.04 SD and -0.12 SD in language (s.e=0.092) and WSD offers a bonus and training to the Evans (2011) math, respectively. principals, teachers and representatives of the primary school community. Kenia Glewee et Effect of 0.067 SD in test scores in the first year, 0.15 SD in the Program offering rewards to primary al.(2010) second year and 0.098 SD in the third year. school teachers based on the test scores of their students, with punishment for not taking the test. India Muralidharan The students who completed primary school with a teacher who Program paying a bonus to rural primary (2011) benefited from the performance-based payment program display school teachers based on the average improvements relative to the control groups by 0.54 SD and 0.35 improvement of their students during 5 in mathematics and language, respectively. years. India Muralidharan Effects of 0.27 SD and 0.17 SD, respectively, in math and Program providing a payment based on & language tests two years after the end of the program. the teacher’s performance: collective Sundararaman bonus based on the school’s performance (2011) and individual bonus based on the teacher’s performance. Beca 18 Developing Glewwe et al. There are 4 studies that evaluate the impact of receiving a merits- countries, (2015) based scholarship on the test scores (in primary and secondary 1990-2014 schools, no evidence in higher education). There are 6 estimates, 5 of which are positive and significant and only 1 is negative (and not significant). Source: WB. 30. The analysis incentive programs show mixed evidence. Organizing the classrooms based on the students’ performance has a positive impact but the same does not go for the organization of schools. The incentives have had interesting impacts. Finally, there is not much solid evidence of impacts of university scholarships. Though there are studies on scholarships for primary and secondary schools that show positive impacts. Analysis of Randomized Experiments, McEwan (2014); Improving Educational Outcomes in Developing Countries: Lessons from Rigorous Evaluations, Murnane and Ganimian (2014); What Really Works to Improve Learning in Developing Countries? An Analysis of Divergent Findings in Systematic Reviews, Evans and Popova (2015); Improving School Education Outcomes in Developing Countries: Evidence, Knowledge Gaps, and Policy Implications, Glewwe and Muralidharan (2015). 156 31. Pedagogic interventions, on the other hand, have a strong positive impact on students’ learning outcomes. The high impact of pedagogic interventions in the world, particularly in developing countries, is one of the most outstanding empirical regularities found in literature about “What works in education?” Clearly, the interventions related to pedagogic coaching or support are the ones with the deepest impact, followed by interventions related to the procurement and distribution of educational materials (see Table 5-2). Table 5-2 Possible impacts of pedagogic interventions Country, Period Evaluation Finding Comments Complete Day in Secondary School Chile Bellei (2009) The effects on students’ learning are positive and The evidence suggests that the program had a significant. In language between 0.05 and 0.07 SD deeper positive impact in rural students, the and in mathematics between 0.00 and 0.12 students who go to public schools and the students situated in the upper part of the distribution of achievement Pedagogic support/coaching. Hiring of pedagogic coaches. Training of trainers Developing Glewwe et al. 11 studies analyze the effect of training active countries, (2014) teachers, with 29 estimates: 18 find positive 1990-2010 effects, but 7 are insignificant. On the other hand, 11 find negative effects but only 1 is significant Procurement and distribution of educational materials Developing Glewwe et al. 21 studies analyze the effect of educational countries, (2014) materials. Total of 60 estimates, 36 of which find 1990-2010 positive effects (26 significant and 10 insignificant). And 17 find negative effects (4 are significant) Kenia Glewee et Impact of 0.02 SD on the learning of all students The limited impact on all students is caused by the al.(2009) (s.e=0.102) and in the highest performance fact that the government texts are very complicated quintile, the impact is 0.22 SD (s.e=0.097) for the average child Developing Krishnaratne et Based on 5 studies, a combined impact size is The program provides more educational materials countries al. (2013) determined of 0.284** SD in language to the teachers Sierra Leone Sabarwal et al. No impact is found in learning improvements The authors argue that the children do not have (2014) access to the books that are stored Interventions in Bilingual Intercultural Education South Africa Sailors et al Positive and strong effects are found for the home (2010) language initiative. At 12 months, the impact found is 0.23 SD in language and at 24 months 0.58 SD in language Source: WB. 32. Pedagogic support for teachers promises the best educational outcomes . But on this, it is important to be cautious because: (i) these results take time to mature, and (ii) the required investments are high. The ideal situation for the educational system would be that in every new generation the most talented youth choose a career in education. This requires high investments in a sustained manner to ensure that teaching is an attractive option for the best students. 33. In the last four years, the expenditure on high-impact pro-poor interventions have increased. Regarding the interventions identified in classification of the expenditure, investments have increased from 0.07 percent of GDP to 0.26 percent between 2012 and 2015. At the same time, the only highly-effective intervention with a “neutral” impact on the poor is the one related to educational material, which reached a peak in 2013 to then remain constant relative to GDP (see 157 Figure 5-21). Below we will see that in the case of Peru, the expenditure on materials has indeed been pro-poor. On the other hand, it was not possible to adequately identify the low-impact spending in the SIAF data, except for two pro-poor interventions: support for BIE and enhanced coverage of basic education. Investments in empirically effective interventions do not ensure that those interventions will have the expected impact, since this depends on the specific design and the quality of implementation, though it is a step in the right direction (see Table 5-3). Table 5-3 Impact and potentially scalable interventions by taxonomy of the effectiveness Low effectiveness - Pro-Poor High effectiveness - Pro-Poor •Juntos •Public Teaching Career •Rural Centers for Alternate Training (CRFA) •Wage bonuses for teachers in vulnerable contexts (Amazon, •Complete School Day (JEC) border, VRAEM, rural, BIE). •Bilingual Intercultural Education (BIE) •Multigrade pedagogic coaching, BIE •Teacher training •Semáforo Escuela •Reports of the Census Evaluation of Students •Beca 18 Low effectiveness - Neutral impact on poor High effectiveness - Neutral impact on poor •Aprender Saludable •Educational materials ** •High-Performance Schools (COAR) Source: WB. Note: This classification is based on the design of the interventions and their impacts in other contexts. We do not use evaluations of the programs themselves (which do not exist in various cases). Based on international evidence, depending on the initial availability and the complexity of the materials, the impact can be more favorable for the less poor. However, as we will see later in this chapter, in the case of Peru the contrary seems to be true. 158 Figure 5-20 Evolution of the expenditure on highly-effective interventions, 2012-15 (% of GDP) 2012 2013 2014 2015 0.26 0.16 0.12 0.09 0.09 0.09 0.07 0.06 0.03 0.02 0.02 0.02 L O W E F F E C T I VE N E SS – P R O H I G H E F F E C T I V E NE SS – H I G H E F F E CT I V E NE SS – P R O P O O R : I B E & E X P A NSI ON O F N EU TR A L I MPA CT O N TH E P OOR C OVER AG E P O O R : E D U CA T I O NA L MATER I AL Source: SIAF, BCR. Own elaboration. Note: Public Teaching Career, training for active teachers (including training, specialization, training in service, and the programs focusing on Pedagogic Coaching, Pedagogic Support, Improved Rural Secondary School, friendly Schools), Student Assessments (assessments of censuses and samples, reports to families and communities, progress maps and curriculum management as the basis for the tests), and scholarships (including Beca 18 in 2015 and other scholarships and educational credits in previous years). Does not include bonuses for teachers in vulnerable schools or Semáforo Escuela since these could not be identified clearly in the data. 34. A large part of learning improvements can be attributed to better access to learning materials. According to Castro (2016), between 30 and 50 percent of the learning improvement in these schools of the country can be attributed to the improvement of two primary educational inputs: the students’ access to learning materials and the teachers’ access, though to a lower extent, to pedagogic coaching. In the rural areas, these educational inputs explain a larger percentage of the learning improvements. The investment in educational material (available on time) could be pro-poor (see Table 5-4). 159 Table 5-4 Blinder-Oaxaca decompositions of the change in learning between 2011 and 2014 Explained Component by the children aged visits received since Average number of Average number of Public expenditure education received pedagogic support years of preschool (regional average) between 6 and 7 Infrastructure on education Materials Result Unexplained years 2008 Component Percentage of students with a satisfactory 23.4%*** -2.6% 1.7%* 2.4% 5.5% 69.4%*** performance in reading comprehension Percentage of students with a satisfactory 26.2%*** -4.1% 1.8%* 3.1% 7.8%* 64.2%*** performance in mathematics Standardized score in reading comprehension 20.8%*** -2.0% 1.6%** 1.3% 4.7% 72.5%*** Standardized score in mathematics 35.2%*** -3.6% 2.6%*** 1.3% 6.2% 58.2%*** Source: Adapted from Castro (2016), Table 3, Panel B. *** p<0.01, ** p<0.05, * p<0.1, based on significance of the individual contributions. Note: All estimates consider fixed effects at the level of the school and on an annual basis (with control for common temporary variations in all schools). Furthermore, there is control for factors such as the household spending in education, the student’s mother tongue, and the school’s geographical location (urban or rural), even though they lack statistical significance under the proposed specification. Conclusions 35. It is critical to prioritize the education spending by region according to equity criteria and specialized region-specific strategies to increase public spending efficiency and effectiveness (see Table 5-4). This analysis identifies several groups of regions, but a more in- depth and qualitative review would help explain this classification more exactly. The first group is the regions that need to make efficiency adjustments before seriously expanding the expenditure on education. These regions are primarily Madre de Dios and Tumbes. On the other hand, there is a group of regions in which expectations are that an increase of expenditure would improve the quality in the short term since they are already relatively near the efficiency frontier (for example, Amazonas). Ucayali is an interesting case because thanks to its efforts and results in primary education it is the leader in terms of net efficiency in this level and it seems to be able to achieve the same in the other levels where its efficiency is very low. 36. The planning and execution of investments require significant improvements. An important part of the investment budget, particularly in infrastructure projects, is transferred throughout the year without much planning. This limits the design, approval and execution of the projects in the fiscal year, with an impact on the rate of budget execution of the subnational governments. Approval of a National Plan for Educational Infrastructure (PNIE) should contribute to clear cost-efficiency criteria, maximizing the number of students benefited per invested Nuevo Sol and balancing this with equity criteria so as to reach out to the most vulnerable population groups. This prioritization would enable multi-annual planning and collection of the necessary information for preparing investment projects, with a territorial focus to minimize the transaction costs as much as possible (World Bank, 2016). Using up-to-date and standardized data on the infrastructure at the level of school premises, the PIPs could be prepared faster and execution thereof could be monitored in real time through the National Information System of Educational Infrastructure, which is proposed in the PNIE. It is also important to bear in mind the maintenance 160 and operational costs (e.g. the ones related to power consumption) when making investments in expensive items (e.g. elevators and pools). With a higher investment, the share of the current expenditure on maintenance and operation would go up, which should be taken into account in the spending plans. 37. Furthermore, it is important to reallocate pro-poor expenditures to high-impact programs. In recent years, a variety of programs has been created and discontinued aimed at improving the quality of education. One criterion for decision-making on the programs is the taxonomy of effectiveness and the pro-poor impact proposed in this chapter. This allows to identify policies with a greater impact potential for all, but primarily the poor, in the northeastern quadrant. The classification is based on international evidence and has to be contrasted with local program evaluations, not only in terms of effectiveness and impact, but also in terms of cost-effectiveness. 38. Improving the information provided for a better schooling management seems to be the most cost-effective option, though it does not solve everything. Thus, efforts such as the “Semáforo Escuela”, the ECE primary and secondary school performance reports and portals such as “Ponte en Carrera” for higher education, offer considerable potential. It would be very helpful for these information flows to indicate, first of all, the different considerations in the educational system and to maintain a close link with the Performance Commitments. But beyond this, it is important for these flows to reach the entire educational community, particularly students and parents. However, the “Semáforo Escuela” reports themselves (http://www.minedu.gob.pe/semaforo-escuela/) show that the UGELs with the highest indices in terms of the non-attendance of school principals, teachers and students are the ones located in the forest areas that are very rural areas, where information and monitoring do not address the underlying problem (distances, costs and traveling times). 39. Another important suggestion is to give priority to the work with teachers; despite the high cost of this type of investments, it is the one with the greatest potential impact. This is especially true for sustainability of the impacts in the long term. Moreover, the investment in teachers is essential, beyond the learning that can be measured in standardized tests. An effective teacher could have a positive impact on his students in a broad range of relevant results (Jackson, 2016). Accordingly, one recommendation would be to prioritize consolidation of the Public Teaching Career, including the wage bonuses for teachers in vulnerable contexts (Amazon, border, VRAEM, rural and BIE), and emphasis on the training and focused coaching for the teachers and principals. 40. In the medium term, the introduction of per capita financing mechanisms as part of a more integral reform of the systems of fiscal transfers to subnational governments, may contribute to higher spending efficiency, equity, predictability, and accountability. The introduction of per capita financing formulas may ensure that adequate human resources are assigned to the schools and that the criteria of horizontal and vertical equity are complied with (Romaguera & Alcázar, 2011). But in order to implement a formula of this type in Peru, at least four preconditions that are not met today would have to be put in place. First, this should be done in the context of a general/comprehensive reform of the system of fiscal transfers to the subnational governments with the objective of avoiding significant alterations of the proportions of the fiscal pie divided among the different functional sectors and specific functions. Second, it would be 161 necessary to periodically have reliable and detailed information about the selected indicators in an integrated information system with a single student identification (e.g. DNI). Third, it would be necessary to have clarity regarding the roles and responsibilities of different government bodies of the sector, especially the schools and their levels of autonomy in budget execution, which requires a greater administrative capacity. Fourth, in the rural area it would be necessary also to set up school management networks that group nearby multi-grade schools under the same expenditure unit, with a fulltime network director and the necessary administrative support. This would entail eliminating the position of principal for the schools that are part of a network. 41. Table 5-5 gives an overview of a series of policy options to enhance spending efficiency and equity in education. The suggested measures and options may give rise to fiscal savings in the medium term, which would be difficult to quantify in the short term because of their nature. These savings that may be obtained eventually could help finance the growing pressure for more investment, access and quality in the sector. Table 5-5 Options to enhance the efficiency of expenditure in the education sector Measure Efficiency Equity Ensure adequate financing for the current expenditure required for the provision and supervision of educational services in different + + contexts, and integrate it into the results-based budgeting (RBB) Allocate a multi-annual investment budget, in line with the priorities + Neutral and strategies of the National Plan for Educational Infrastructure (PNIE), including the necessary operational expenses Prioritize the expenditure on highly-effective and pro-poor + + interventions and policies Implement an electronic government that connects the schools and optimizes the information flows, reduces the transaction costs and + + speeds up administrative procedures Draft differentiated strategies (by region, university) and accredit capacities at the level of the UEs and the schools in order to increase + + budgetary responsibilities Put in place the conditions to allocate the subnational budget through student-based formulas, including equity criteria, starting + + with the accredited schools and UEs. Link the budget to performance-driven incentives + Neutral Source: WB. 162 Introduction 1. Citizen security is one of the sectors in which spending increased the most between 2010 and 2015. The expenditure on citizen security almost doubled in real terms between 2009 and 2015. This sector accounts for around 9 percent of the national budget and ranks fourth in terms of the total spending after education, health and transportation in 2015. Even though the level of violence is relatively low in Peru compared to the region, the levels of violence have tended to go up coincidently with the intensification of drug trafficking. The effective fight against crime and violence does not necessarily mean spending more but rather spending better and using the available resources in an optimal manner. In effect, Peru’s efficiency of its expenditure on security is low relative to other countries. At the same time, the security sector faces serious challenges, such as the use of outdated technology and the absence of basic services to operate in half of the police stations. In this context and at a time of fiscal consolidation, it is especially important to improve the spending efficiency in the short and medium term at this time of fiscal consolidation in Peru. 2. The reallocation of resources for the national police may give rise to spending efficiency gains of more than 0.3 percent of GDP. The objective of this chapter is to examine the expenditure on citizen security and measure the efficiency thereof. The chapter seeks to analyze where and how the money is spent and whether the recent increase in the public expenditure on citizen security has had an impact on the levels of crime and violence, and what the potential gains are in terms of efficiency. When compared internationally, Peru’s spending in citizen security is relatively high, but the levels of efficiency are relatively low. For example, on the district level the national police (PNP) has efficient results in only 8 to 11 percent of the districts. In this sense, reassigning policemen and inputs with a view to optimizing the input-output ratio may give rise to a significant impact in the service while at the same time generating significant savings. The potential savings could be greater when taking into account the efficiency of the police organization as a whole. 3. The chapter is structured in five parts. The first part briefly examines the context in terms of (in)security in Peru over the last decade. There is also an analysis of the public spending in citizen security and of the factors that determine the expenditure in citizen security. The second part presents the institutional structure and the overall challenges for efficiency in the aggregated spending. The third part examines the levels of efficiency of the National Police of Peru (PNP) as the institution responsible for fighting crime and violence and providing citizens with security. This is done using the frontier analysis methodology (DEA) which estimates the optimal combination of inputs for production. Furthermore, the study estimates the potential gains in terms of efficiency if the inputs were reduced or reassigned. With this methodology, it is possible to see how optimizing the cost structure of the used inputs may lead to the possible maximum production at a minimum cost. The fourth part presents the results of the original analysis of the impact of a greater efficiency of the PNP at the social level in terms of a reduction of victimization and the perceptions concerning the PNP’s performance. For this analysis, the study uses the quasi- experimental method of propensity matching (also see Annex 4). The fifth part presents a series of reform options to enhance efficiency of the spending in citizen security. 163 Expenditure and Results Expenditure 4. The expenditure on citizen security has increased considerably since 2008. In nominal terms, it almost doubled between 2009 and 2015 and it has increased by 41 percent since 2012. Practically the whole increase in the sector of order and security in 2012 was explained by higher wages of the police personnel, the increase of which was equivalent to 0.16 percent of GDP. In the same period, the goods and services in the sector of order and security increased by the equivalent of 0.03 percent. 5. At the international level, Peru spends a lot more than its peers relative to its total budget but not relative to Peru’s GDP. Peru spends around 9 percent on citizen security relative to its total budget (see Figure 6-1). The average expenditure on citizen security in OECD countries is approximately 4 percent of the total budget; the average of its structural peers is 6 percent. Furthermore, this high expenditure has remained constant throughout the period covered by the study. Relative to GDP, Peru has a similar expenditure (1.6 percent) compared to the OECD average and the average of its structural peers, which is also 1.6 percent of GDP. 164 Figure 6-1 International comparison of public spending in citizen security Expenditure relative to the total budget Peru Pares 8 6 4 2 0 Spain Germany Greece Sweden Peru Poland Peru United States Slovak Republic Poland Romania Norway Portugal Korea, Rep. Finland United Kingdom Chile Iceland Netherlands Ireland Pares estr. Pares OECD reg. Expenditure relative to GDP Peru Pares Media 2.0 1.5 1.0 0.5 0.0 Romania Poland Australia Chile Canada Peru Chile Peru Portugal United Kingdom Slovenia United States Spain Slovak Republic Netherlands Italy Poland Australia New Zealand Chile Israel Switzerland Peru Ireland Iceland Hungary Greece Sweden Japan Belgium Estonia France Czech Republic Norway Germany Korea, Rep. Finland Austria Denmark Pares Regionales Pares Estructurales OCDE Source: WB analysis based on OECD data of 2015 (2016). Note: the general government expenditure includes the expenditure on the three government levels (central, regional and municipal). 6. Expenditures on citizen security have increased significantly in recent years, especially at regional and municipal levels. The public spending on security at the regional and local levels increased most between 2010 and 2013 to then decline significantly in 2014 and 2015. In contrast, central spending increased more gradually in the same period and despite a fall also in 2015, the central government carried the weight of the expenditure on citizen security (see Figure 6-2). 165 Figure 6-2 Evolution of real expenditures by administrative government levels 80 60 40 20 0 2010 2011 2012 2013 2014 2015 -20 -40 Central Local Regional Source: WB analysis based on BOOST-World Bank (2016). Values in million PEN. 7. The current and capital expenditures both contributed to the rise in citizen security spending. Even though current expenditure accounts for the highest percentage of the budget (80- 90 percent), the real variation in evolution of the expenditure shows an effort of investing in capital expenditure, particularly in public infrastructure works in order to adapt the infrastructure of the police stations so they could perform their work and provide services to citizens and victims (see Figure 6-3). For example, the national census of police stations (CENACOM) shows that only 53 percent of the police stations covered by the census at the national level have basic services to operate (potable water, sewerage, power and permanent connection to the public network). Furthermore, in 2014 the current expenditure was increased as the result of a package of economic measures to improve the wages of the police force. 8. Per capita expenditures on citizen security at the subnational level have increased significantly in recent years, though there is considerable variation among regions (see Map 6-1). This expenditure has increased progressively in the period covered by the study from 165 PEN per capita in 2009 to almost 300 in 2015 (299 PEN per capita 47). However, this increase has not been homogeneous in all departments. While the departments of Lima or Cusco have an average per capita expenditure of more than 400 PEN, departments like Huancavelica or Amazonas have spent an average of 47 and 54 PEN per capita, respectively. 9. One of the basic principles of public financial management is that the allocation of resources is in line with the needs of the sector. Citizen security is one of the priorities of the Government of Peru. According to the National Citizen Security Policy, “constant improvement of the levels of security of the Peruvian population is one of the priorities of the State” since security contributes to development and prosperity of the country. The strategies in citizen security consider that citizen security is a cross-sectoral topic with a focus on the local level and long-term prevention besides ensuring the control and reduction of violence and crime. 47 The value is at constant prices. 166 Figure 6-3 Public expenditure on citizen security, by economic classification 60 10000.00 50 9000.00 8000.00 40 7000.00 30 6000.00 20 5000.00 10 4000.00 3000.00 0 2010 2011 2012 2013 2014 2015 2000.00 -10 1000.00 -20 0.00 Current Spending Capital Spending Total Source: WB analysis based on BOOST-World Bank (2016). Values in million PEN. Map 6-1 Per capita spending in security at the departmental level Source: WB analysis based on BOOST-World Bank (2016). Results 10. Security is a basic public good for Peruvians and a key pillar for economic growth and social development. Citizen security is a priority for the Government of Peru and Peruvian society. In the last 20 years, this problem has been addressed through different approaches with the objective of reducing violence and promoting economic growth, social development and democratic consolidation of the country. 167 11. Despite efforts to reduce violence, the level of violence has been increasing in the last five years. The levels of violence reflected in the rate of homicides per 100,000 inhabitants have experienced a rebound at the national level over the last five years and, in some departments, the number of homicides has increased quite significantly. For example, the rebound of homicides in the departments of Tumbes and Madre de Dios coincides with the increase of drug trafficking in recent years and has had an impact on the rate at the national level (see Map 6-2). Map 6-2 Evolution of the homicide rate per 100,000 inhabitants at the departmental level 2011 2014 Source: WB analysis based on data of INEI (INEI, 2016). 12. Nonetheless, the levels of violence are relatively low compared to other countries in Latin America. The rate of homicides in Peru is among the lowest compared to the regional level. The average rate of homicides between 2000 and 2013 in Peru is the second lowest in the region after Chile and is significantly lower than the regional average (21.80) in the same period (see Figure 6-4). 168 Figure 6-4 Average rate of homicides per 100,000 inhabitants at the regional level (2000- 2013) Source: WB analysis based on data of UNODC (UNODC, 2014) 13. The crimes against property such as theft and robbery with violence have increased rapidly in recent years. According to data reported by the PNP, the crimes against property are the ones that have increased the most in terms of the number of reports filed, from 108,000 reports filed in 2009 to 185,000 in 2014. This number is significantly higher than other crimes, e.g. the offences against life only increased from 20,000 reports filed in 2009 to almost 28,000 in 2015. Other important offences, such as offences against freedom, have remained relatively stable between 2009 and 2015 (INEI, 2015). 14. Regional public opinion surveys on victimization confirm these trends and provide additional insights. Regional public opinion surveys such as Latinobarómetro and LAPOP show a slight decrease in the levels of victimization among the Peruvian population 48. For example, the Latinobarómetro survey shows a progressive decrease of the number of respondents that were victims of violence. In the 1990s this number was almost 50 percent while in the most recent survey in 2015, this number went down to only 10 percent. The ENAPRES survey shows that the national rate of victimization in urban centers declined from 40 percent in 2011 to 29.5 percent in 2015.49 15. Overall, Peru’s level of homicides by international comparison is in line with its underlying country characteristics. Invariably, socioeconomic and demographic factors have an impact on violence and crime (World Bank, 2011a, 2011b). The factors identified in literature as the ones with the greatest impact on the levels of violence and crime are: inequality, poverty, the percentage of youth in the population, and rapid urban growth, among others (World Bank, 2010). Figures 1 to 5 in Annex 4 show the dispersion linking several socioeconomic and demographic variables identified in literature to the rate of homicides per 100,000 inhabitants. The results show 48 It is important to underline that the values may vary because of the type of sample used in the different countries. 49 These are raw data that have not been weighted or adjusted by strata. Therefore, they have to be interpreted carefully. 169 the global trends and the performance of Peru in the correlations between socioeconomic factors and the rate of homicides50. 16. These crimes also have an important impact on the economy and firms incur substantial costs to protect themselves from violence and crime. The World Bank Enterprise Surveys show the alarming economic impact of violence and crime in the private sector. Almost 80 percent of the companies make investments in security, which is significantly higher than the average rate among structural peers 51, regional peers52 and the OECD countries. Likewise, the spending in security and the percentage of losses from crime is higher in Peru compared to its peers and OECD countries. Finally, the survey shows that 25 percent of the companies in Peru assert that they suffer losses due to crime. This percentage is lower than the average among its peers and OECD countries (see Figure 6-5). Figure 6-5 Costs of violence in the private sector – Peru 90 80 70 60 50 40 30 20 10 0 Firms that pay for Firms with loses due Expenditures on Losses as a result of Security Costs (% of security (%) to crime (%) Security (% of crime (% Annual Total Sales) Annual Sales) Sales) Peru Structural Pairs Regional Pairs OECD Source: WB analysis based on the World Bank Enterprise Surveys (2016). Note: World Bank. 2012. “World Enterprise Survey: Peru.” Consulted on 1 May 2016 at http://www.enterprisesurveys.org/data/exploreeconomies/2010/peru. The data for the countries included in the analysis are for different years. 17. Institutions such as the PNP have the fundamental role of providing citizen security, protecting citizens and combating crime and violence. The survey results show that the population does not trust the PNP to solve insecurity issues. If the police institutions do not respond or respond in a limited manner to the population’s needs, probably the levels of crime and delinquency will go up as well as the levels of insecurity in the population. 50 The rate of homicides is used as the indicator that is most robust to underreporting and the one that is most comparable at the international level. 51 The structural peers of Peru are Australia, Canada, Chile, Colombia, Malaysia, Poland, Romania and Thailand. 52 The regional peers of Peru are Chile, Colombia and Mexico. 170 18. In the last decade, the issue of insecurity has become one of the greatest concerns of the population. At the beginning of the last decade, citizen security was put on the discussion agenda. The result was a state policy aimed to eradicate violence and strengthen citizen security (Dammert, Ruz, & Salazar, 2008). The result of this policy is creation of the new National Citizen Security System (SINASEC) through Law No. 27933 (cite). The SINASEC is responsible for coordinating state actions related to citizen security while at the same time promoting citizen participation (SINASEC, 2016). 19. The insecurity caused by crime and violence creates distrust in the population towards institutions such as the PNP. The Latinobarómetro survey and the National Survey of Strategic Programs (ENAPRES) show that 9 out of every 10 people (91 percent) who live in urban centers in Peru do not trust the PNP. Peruvians’ opinion about the police’s performance is not encouraging either. Between 60 and 70 percent of the respondents have negative opinions about the PNP’s performance in terms of providing prompt information to citizens, ensuring security, treating people with respect or keeping the community informed about citizen security (INEI, 2015). 20. The increasing insecurity combined with the need for addressing crimes with a multi- sectoral approach has given rise to the creation of a complex citizen security system that seeks to ensure coordination of different sectors from the three levels of government. Peru’s citizen security system is multi-sectoral and multi-level. It is made up of ministries from the social, transportation and communications sectors, as well as the ministries of the criminal justice system.53 This complex system of citizen security is led by the National Citizen Security Council (CONASEC) and the state is the guarantor of security together with the regional and district governments, with coordination among the three government levels. The coordination difficulties sometimes give rise to insufficient financing or deficient budget execution. The next section examines the spending in citizen security, the evolution thereof and execution on the three levels of government. Moreover, it measures where the expenditure has had an impact on crime and violence. Efficiency and Effectiveness of the Expenditure Institutional structure 21. The primary objective of the Citizen Security Plan 2013-2018 is to satisfy the needs concerning citizen security from an intersectoral perspective. This plan systematizes the contributions of the institutions making up the SINASEC, of civil society and of the private sector and seeks to reflect the regional and local, urban and rural diversity and reality of Peru. 22. The Citizen Security Plan proposes to reduce the violence and crime on various levels that affect insecurity and the public spending aims to finance the institutions of the sector. On the one hand, the objective of the Plan is to reduce the rates of crime and violence. On the other hand, it seeks to improve effectiveness of the actions to combat crime and violence. Furthermore, the Plan considers the reduction of violence against women, the reduction of the percentage of 53 For more information about composition of the cross-sectoral citizen security committee, please see the following document: http://www.descentralizacion.gob.pe/wp-content/uploads/2015/03/03.07.CONASEC.pdf 171 persons deprived of liberty and the effective fight against drug use and trafficking. As a result of the axes of the Citizen Security Plan, the public expenditure therefore seeks to finance the institutions responsible for controlling, reducing and preventing crime and violence. 23. The public spending in security comprises seven primary programs, of which the PNP program absorbs 54 percent of the total spending. Of these seven programs, the most important one is 0030—a reduction of crime and failures affecting security—which is the one with the highest budget. The amount allocated to this program is an average of 80 percent of the public spending budget in security.54 The largest part of the budget of program 0030 (65 percent) is allocated to police operations that are a critical element in the deterrence, prevention and reduction of crime and violence. Since 2012, the budget allocation of these security programs has increased by 65 percent. Moreover, the programs on the fight against terrorism (0032) and illicit drug trafficking (0031) (see Table 6-1) have grown rapidly. Table 6-1 Budget Programs in Citizen Security Programs 2012 2013 2014 2015 0030. REDUCTION OF THE VICTIMIZATION IN URBAN ZONES OF THE COUNTRY 3,068 3,588 4,670 4,771 WITHIN THE FRAMEWORK OF CITIZEN SECURITY 0031. REDUCTION OF ILLICIT DRUG TRAFFICKING 157 187 208 271 0032. FIGHT AGAINST 82 501 390 435 TERRORISM 0048. PREVENTION OF AND RESPONSE TO FIRE, MEDICAL 32 123 104 59 EMERGENCIES, RESCUES AND OTHERS 0068. REDUCTION OF VULNERABILITY TO DISASTERS 93 302 383 551 AND DISASTER EMERGENCY CARE 0086. IMPROVEMENT OF THE SERVICES OF THE CRIMINAL 0 269 21 285 JUSTICE SYSTEM 0128. REDUCTION OF ILLICIT 0 0 0 28 MINING Grand Total 3,433 4,971 5,776 6,400 Source: BOOST-World Bank (2016). Note: Program 0139 “Decrease of the Incidence of violent social conflicts, protests and mobilizations that disturb the Public Order” only has data as from 2016. The PIM amount in 2016 for this program is 255,857,137 soles. Efficiency of the public budget 24. There is no specific formula that defines how to allocate resources to citizen security. According to Shack et al. (2013), the allocation of resources depends on various factors and, 54 The data are available only as of 2012 because before the budget was not by program but only by function. Therefore, the evolution of the budget by program can be evaluated only as from 2012 to date. 172 according to the authors, the levels of crime and violence, and the total population are the two factors that had the greatest influence on the resource allocation. The authors also argue that there is considerable heterogeneity in the concentration of crimes and the allocation of resources has been concentrated in the departments with most crimes, but there has been no increase in the budget allocation to the department where delinquency has increased. 25. Socioeconomic and demographic factors have a stronger correlation with the budget allocation than the number of crime reports filed. The results of the original regression analysis show that the budget allocation is influenced by socioeconomic and demographic factors. The population and the total number of PNP officers continue to be strongly associated to the budget allocation. The total number of reports filed is associated positively to the budget allocation but it is not significant from a statistical point of view. The level of inequality and the poverty rate are also associated to the budget but with opposite signs. Inequality has a positive sign and poverty a negative sign. Possibly, these are capturing other factors related to violence and crime (see Table 6-2). Table 6-2 Regression Analysis of Budget Allocation Variables (2013) Total Budget (log) Total Budget (log) Population (log) 0.206* 0.404*** [0.10] [0.11] PNP personnel (log) 1.148*** 0.884*** [0.13] [0.14] Reports (log) 0.0780 0.00979 [0.06] [0.06] Illiteracy Rate 0.00771 [0.01] % youth aged 15 - 24 -0.0237 [0.02] Inequality Coefficient 6.787*** [1.99] Unemployment Rate -0.000465 [0.07] Poverty Rate -0.0177** [0.01] Constant 6.484*** 4.687*** [0.69] [0.93] 2 R 0.408 0.426 Observations 584 583 Source: WB analysis based on data of CENACOM, RENAMU, INEI (INEI, 2016). Note: Standard errors between square brackets, * p<0.05 ** p<0.01*** p<0.001. The analysis is conducted for year 2013 as the only year with data broken down by district at the level of crime and socioeconomic and demographic information. 173 26. Expenditure are relatively high relative to the total budget but budget execution is relatively low, especially at the regional and local levels, despite considerable improvements in recent years. Budget execution has improved substantially over the last three years, both in terms of overall execution and in terms of the execution of different administrative levels of the government, but the percentage of execution of 70 percent at the regional and local levels is still low (see Table 6-3). It is important to note that there is considerable variation in execution at more disaggregated administrative levels, depending on the type of programs and/or activities conducted. Table 6-3 Execution of expenditure on citizen security Year 2009 2010 2011 2012 2013 2014 2015 Peru Modified Amount (Real Prices) 5397 6233 6547 7660 8452 9995 10359 Executed Amount (Real Prices) 4490 5098 5039 6669 7608 9090 9439 Percentage of Execution 83% 82% 77% 87% 90% 91% 92% Central Modified Amount (Real Prices) 4612 5330 5367 6136 6640 8115 8574 Government Executed Amount (Real Prices) 4221 4718 4656 5595 6206 7677 8174 Percentage of Execution 92% 89% 87% 91% 94% 95% 95% Regional Modified Amount (Real Prices) 154 261 360 408 422 315 311 Government Executed Amount (Real Prices) 76 125 121 234 335 236 214 Percentage of Execution 49% 48% 34% 57% 79% 75% 69% Local Modified Amount (Real Prices) 631 643 821 1115 1389 1565 1474 Government Executed Amount (Real Prices) 194 256 262 840 1067 1178 1051 Percentage of Execution 31% 40% 32% 75% 77% 75% 71% Source: WB analysis based on BOOST-World Bank (2016), million soles. 27. The low budget execution at the subnational level is related to coordination and planning issues, as well as the constraints of the information system to identify spending targets. There are studies identifying various factors that could explain the low budget execution at the local and regional levels. As in other sectors, the ministries first maximize the functional budget. Subsequently, the ministries have an incentive to maximize their transfers to the local governments so that they end up with high execution ratios, transferring the apparent inefficiency in execution to the subnational governments (see the macro-fiscal chapter). In addition, there are shortcomings in the planning and coordination between the PNP and the Ministry of the Interior (MININTER), particularly between the technical area of the MININTER and the PNP. This is reflected in, for example, the lack of a budget in accordance with the needs of the PNP for the maintenance of infrastructure and vehicles. Moreover, there is a disjunction between the information systems and the program execution targets, which are difficult to set since the information systems are not very reliable. Finally, the limited coordination among the three administrative levels for budget allocation may affect budget execution (Ministry of Economy and Finance, 2013; Shack et al., 2013). 28. The increase of the expenditure on citizen security at the subnational level was negatively correlated with the levels of violence between 2012 and 2014. The preliminary analysis of the impact of the expenditure on citizen security shows that, on average, higher spending levels at the departmental level coincide with a decrease of the rate of homicides, albeit 174 that the impact has been very limited in both. Figures 6-6 and 6-7 show the impact of an increase in the expenditure in million PEN, the rate of homicides per 100,000 inhabitants and the rate of victimization at the departmental level for years 2012 to 2014. For example, an increasing investment in citizen security from 5 to 10 million PEN in a department with a rate of 5.3 homicides per 100,000 inhabitants is associated to a reduction in that rate to 5.2 homicides per 100,000 inhabitants. Figure 6-7 shows that the effect of the expenditure on the levels of victimization increases with values below the average expenditure on citizen security for the period covered by the study (268 million PEN) to then fall to levels above the average for the period covered by the study. For example, the counterfactual simulations to make the estimates show that the increasing expenditure on citizen security from 30 to 40 million PEN in a department was associated to an increase of victimization by 0.07 percent, while an increase of 300 million corresponded to almost 5,000 in the department of Lima, with a reduction of 0.05 percent in the rate of victimization. Figure 6-6 Impact of spending on Figure 6-7 Impact of spending on homicides victimization Source: WB analysis based on data of CENACOM, RENAMU, INEI (INEI, 2016). Note: The horizontal axis shows the expenditure on citizen security at the departmental level; the vertical axis shows (a) the homicides at the departmental level, and (b) the victimization at the departmental level. Spending Efficiency of the National Police of Peru (PNP) 29. Translating the increase of public spending on citizen security into better results requires enhancing the efficiency in delivering these security services. The efficient use of the human, financial and infrastructure resources will not only lead to a greater and better optimization of these resources but also to a deeper impact in the social sphere in terms of reductions of the victimization and insecurity and an improved perception regarding performance of the public institutions that provide citizen security services, such as the PNP. 30. The most important institution in the fight against crime and violence is the PNP. The police bodies as public institutions have a budget that has increased annually because of different reasons, one of the clearest ones of which is the growing demand for more security in view of the growing levels of insecurity in the country. Using the frontier analysis method, it is possible to estimate the levels of PNP efficiency. The efficiency can be estimated by comparing the distance of the PNP production from an efficiency frontier. This frontier shows whether the available resources have been used efficiently for producing the PNP products. This frontier is estimated by using the data envelopment analysis (DEA) method, using the inputs and outputs of Table 6-4. 175 The aim of this methodology is to find the units of a sample with the best performance in terms of the use of resources, which generated a maximum efficiency frontier. The distance from a unity to this efficiency frontier provides the estimated levels of (in)efficiency. For example, if a police station does not efficiently use inputs such as police officers, vehicles or computers, it will be located below the efficiency frontier, compared to a police station that used its inputs efficiently. In accordance with the literature about the efficiency of police organizations (see for example, Barros, 2007; Díez-Ticio & Mancebón, 2002; Gorman & Ruggiero, 2008), the following inputs and outputs are used (see Table 6-4). Table 6-4 Inputs and outputs for the efficiency analysis Inputs Outputs Global analysis # of crimes prevented (using the reverse of the number of # of police officers reports filed as a police output) Departmental and district analysis # of crimes prevented (using the reverse of the number of reports filed as an output that captures the potential for crime # of police officers prevention by the PNP) # of vehicles # of foot patrols by month Wages # of vehicle patrols by month # / percentage of crimes recorded (for the analysis at the # of technology elements departmental level) # of real estate elements Source: WB analysis based on the National Survey of police stations (INEI, 2016) 31. The efficiency measurement captures the PNP’s capacity to prevent crime and violence. In this sense, this study uses the total number of crimes potentially prevented by the PNP as the output of the police activity, as a result of the combination of inputs such as the total number of police officers, vehicles and wages, which amply capture the police work. This is based on the premise that one key element in the police production is the prevention of crime and violence. 32. The analysis is conducted at three levels: national, departmental and district. First, an analysis is made at the international level, comparing Peru to the other countries of the region and OECD countries. The second analysis is made at the departmental level and the final analysis at the district level.55 55 It is important to note that the results in terms of efficiency are confined to the information coming from the police stations which was collected through the census of police stations published by INEI. Therefore, the efficiency is estimated only for the inputs and outputs produced at the level of the police station and not for the entire police body of the PNP. 176 Efficiency analysis at the global level 33. The PNP’s efficiency is low compared to its structural and regional peers. The results of a global study conducted by Parra et al. (2015), using the crime clearance rate as the output and the total number of police officers as the input, shows that the PNP’s efficiency in Peru is significantly lower than the efficiency of the police of its structural peers (81 percent) and of its regional peers (72 percent) such as Chile and Colombia that had levels of efficiency of 100 percent during the study period (see Figure 6-8). Figure 6-8 PNP efficiency on the global level Efficiency Levels Peru Pairs Average 1.20 1.00 0.80 0.60 0.40 0.20 0.00 Netherlands New Zealand Sweden Greece Luxembourg Poland Spain Poland Germany Norway Japan Denmark Switzerland Czech Republic #N/A #N/A #N/A Malaysia Mexico Peru Thailand Peru Peru Peru Austria Finland Canada Canada Estonia Portugal France Slovenia Romania Slovak Republic Colombia Chile Colombia Chile Chile Structural Pairs Regional Pairs OECD Source: Adapted from Parra et al. (2015). Efficiency analysis at the departmental and district levels 34. At the departmental level, four inputs and one output are used to measure efficiency. To analyze the PNP’s efficiency at the departmental level, the inputs used were the total number of computers per department, the real estate elements in the police stations, the technology (computers and telephony) and the number of police officers. The output was the crime clearance rate.56,57 35. The performance of the PNP in terms of technical efficiency is relatively low at the departmental level. The input orientation analysis shows that the average efficiency was 80 percent, i.e. the inputs could be reduced by 20 percent to maintain the same performance. For example, if a department uses police officers, vehicles and technology (computers, telephones, etc.) to do its work and it achieves a level of input-oriented efficiency of 80 percent, the PNP in this department would have to reduce or reallocate 20 percent of the inputs in order to be efficient. On the contrary, the level of output-oriented efficiency shows an average level of efficiency of 40 56 The crime clearance rate is one of the classic outputs of police work. It is calculated by dividing the number of arrests by the total number of crime reports. The definition of ‘crimes cleared’ differs depending on the country in which the analysis is conducted. 57 It is important to note that the efficiency analysis does not capture a series of important aspects of police work such as the quality of the police officers, their experience, the level of discretion and other intangible aspects that are critical for the police production. Nonetheless, the absence of these aspects does not mean that the PNP cannot work efficiently. 177 percent. For example, if a department achieves an output-oriented efficiency of 40 percent, its production could increase to 60 percent, using the same inputs. Correcting for potential biases 58, it is clear that none of the departments were efficient during the study period. The maximum level of input-oriented efficiency was achieved by the department of Moquegua with 92 percent and in the output-oriented analysis the department of San Martin displayed the best performance with 73 percent (see Maps 6-3 and 6-4). Map 6-4 Input Efficiency of the PNP Map 6-3 Output Efficiency of the PNP Source: WB analysis based on the National Census of police stations and Statistics on Citizen Security (INEI, 2011-2015, 2016). Note: Simar Wilson method (2000) with 2,000 bootstrap replications, at the departmental level. 36. The levels of efficiency of the PNP at the departmental level are low compared to other countries in the region.59 Even though few studies have examined the police’s efficiency by departments in Latin America and the Caribbean, they only analyzed the output-oriented efficiency to improve the police production (see Maps 6-5 and 6-6). These studies show that the levels of output-oriented efficiency amount to, for example, an average of 70 percent at the level of the federal entities in Mexico (World Bank, 2015),and an average 60 percent at the departmental level in Guatemala (Alda, 2014). Honduras, however, has an output-oriented police efficiency of only 30 percent (Alda, 2013). 58 Outliers may introduce important biases if they push the other observations below the efficiency frontier. To correct these potential biases, we use the proposal of Simar and Wilson (2000). 59 It should be noted that the input and output variables vary from study to study and may therefore yield different levels of efficiency. 178 Map 6-6 Input Efficiency of the PNP Map 6-5 Output Efficiency of the PNP Source: WB analysis based on the censuses of police stations and the National Register of police stations (INEI, 2016). Note: At the district level. 37. The input-oriented analysis reveals considerable variation in spending efficiencies across police districts. The average efficiency for the input-oriented analysis is 50 percent; the efficiency for the output-oriented analysis is slightly lower at 41 percent. In other words, the PNP could maintain its performance and reorient or reduce its inputs by 50 percent and maximize its production by 59 percent with the inputs used at present. The efficiency scores range from 4 percent for the district of San Juan de Miraflores to 100 percent for the district of San Juan de Lurigancho or Vista Alegre. In fact, only 90 districts out of a total of 795 districts (11 percent) on which data are available, are efficient. Furthermore, the same analysis on the district level, but with an output-orientation, shows that only 8 percent of the PNP police stations at the district level were efficient in the use of their resources. Determinants of the (in)efficiency 38. Several external factors 60 influence the PNP’s levels of efficiency. Socioeconomic and demographic variables invariably influence the PNP’s operations and hence its efficiency in resource use. A second phase of the efficiency controls for these factors to identify whether they contribute to improving or lowering the levels of efficiency. 39. The level of inequality in the district, the presence of burglaries, the unemployment rate, and the misuse of alcohol have an adverse impact on the police efficiency. Poverty and a younger population are positively correlated with PNP efficiency improvements (see Table 6-5). This result is as expected, except for poverty. This result is somewhat surprising because 60 Based on data of the national register of municipalities, in which the respondents identify the main crime problems in the municipality. The two main problems identified were house burglaries and alcoholism. The crime factors in the district may have an adverse impact on police efficiency (Barros, 2007). 179 poverty seems to be associated to lower levels of efficiency, considering that in general the police would have to use more resources for its operations in more difficult districts from an operational point of view. However, it is possible also that the districts with higher levels of poverty have less crime or citizens file fewer reports and so the police use fewer resources for control and investigation.61 40. One important factor to be highlighted is that the presence of Serenazgo in the district reduces the PNP’s efficiency. Half of the municipalities in Peru have a municipal security force called Serenazgo that focuses on patrolling and crime prevention, as well as institutional support for the PNP. In most cases the Serenazgo bodies patrol together with the PNP. The results of the regression analysis show that both the variable on whether the district has a Serenazgo body and the variable on whether it patrols together with the PNP, have a negative correlation with the PNP efficiency. That is to say that the presence of the Serenazgo body in the municipality or the patrolling activities of the Serenazgo with the PNP are correlated to a reduction of PNP efficiency by 12 percent and by 11 percent, respectively. 41. Furthermore, there are qualitative factors, not included in the model, that may have a negative impact on efficiency, e.g. the 24x24 regime. Through this program implemented in the PNP, the police officers were given permission to work at the PNP during 24 hours followed by another 24 hours of work for third parties as a manner to compensate the low wages paid by the PNP. Nonetheless, anecdotic evidence suggests that this program gave rise to the police officers becoming moonlighters who did not devote 100 percent of their capacity to the public service of citizen security. In addition, less than 100 percent of the police officers were present in the police stations on all days of the week. Consequently, the capacity reduction combined with the fact that only half of the officers worked at the service of citizens could be translated into inefficiencies in the use of police officers. 42. The degree of influence of external factors on the police organization shows that not a single district is efficient in the use of its resources. 62 The first phase of the analysis shows that in 92 districts, the PNP had a more efficient performance with the available resources. The results of the second phase that measures the influence of external factors are a lower efficiency in the districts in which the PNP was efficient. In other words, the influence of these external factors is important in the measurement of the efficiency of an institution like the PNP that interacts constantly with society. 61 In effect, the analysis shows that the municipalities with a higher poverty rate are also the ones with a lower crime rate. This correlation is statistically significant. 62 It should be noted that all districts in which the PNP operates were available for the analysis. 180 Table 6-5 Regression analysis of the determinants of PNP (in)efficiency [1] [2] [3] [4] Technical Technical Technical Technical Efficiency Efficiency Efficiency Efficiency Population Density -0.00000412 -0.00000373 -0.00000366 -0.00000323 [0.00] [0.00] [0.00] [0.00] % of Immigrants -0.000564 -0.000696 -0.000978 -0.00107 [0.00] [0.00] [0.00] [0.00] Poverty Rate 0.00255** 0.00236** 0.00179 0.00181 [0.00] [0.00] [0.00] [0.00] Unemployment Rate -0.0065 -0.00541 -0.0013 -0.00364 [0.01] [0.01] [0.01] [0.01] Inequality Coefficient -1.109*** -0.976** -0.915* -0.902* [0.33] [0.32] [0.37] [0.35] % with higher education -0.000908 -0.000878 -0.000619 -0.000587 [0.00] [0.00] [0.00] [0.00] % of youth aged 15 to 24 0.00520* 0.00586* 0.00565* 0.00600* [0.00] [0.00] [0.00] [0.00] % of house burglaries -0.0728* -0.0461 -0.0484 [0.04] [0.04] [0.03] Problems of Alcoholism -0.0477 -0.0462 -0.0472 [0.03] [0.03] [0.03] Presence of Serenazgo in the Municipality -0.129*** [0.03] Serenazgo with the PNP -0.105*** [0.03] Constant 0.688*** 0.744*** 0.823*** 0.789*** [0.10] [0.12] [0.13] [0.12] sigma Constant 0.273*** 0.271*** 0.268*** 0.268*** [0.01] [0.01] [0.01] [0.01] Observations 598 598 598 598 Standard Errors between Square Brackets Source: WB analysis based on CENACOM and INEI (INEI, 2011-2015, 2016) Note: Tobit model with bootstrap replications (250). A Tobit model is used since the dependent variable is limited to 0 to 1 or 0 to 100. 43. To quantify the PNP efficiency gains, it is necessary to estimate the cost efficiency. The cost efficiency is considered the most important for public sector entities since it allows for estimating the total value of the minimum number of inputs needed for a district police station to be efficient. This allows for quantification in terms of the total possible cost saving for all inputs used in the police production, through the reduction or reorientation of inputs. 181 Table 6-6 Optimal values of inputs and monetary quantification in savings Police officers Telephony Computers Vehicles Savings in Inputs 9558 2410 5274 1234 Percentage of 30% 31% 44% 37% Total Savings in PEN 1,472,319,182.5 17,851,020.0 23,381,358.0 236,436,795.0 Source: WB analysis based on CENACOM and INEI (INEI, 2011-2015, 2016) Notes: Cost efficiency model with bootstrap replications (2000) (Simm, 2016). The values are not calculated in real prices. 44. The results of this analysis show that the PNP the district level has very low levels of cost efficiency at the district level; the reallocation of inputs across districts could thus improve citizen security. The results show that the average cost efficiency is only 35 percent, ranging from 2 percent to 100 percent. As a result of these low levels of efficiency, the reduction in terms of inputs is quite significant (see Table 6-6). For example, the PNP could reduce its number of police officers by 30 percent to be efficient. Likewise, the reduction of the number of vehicles is 31 percent. The savings in PEN are considerable. For example, the PNP could save 1.5 billion PEN in terms of wages and 236 million PEN in vehicles to improve its efficiency in use of the resources in the district police stations. 45. These required reallocations can be substantial. Estimates of the optimal number of inputs show that the inefficient police stations would have to reduce or reallocate inputs to be efficient, which could be transformed into fiscal savings. It is important to analyze the impact of the reduction in current PNP expenditures relative to the budget and to GDP—17.4 percent and 0.27 percent, respectively—and of the cost of vehicles—2.8 percent and 0.04 percent, respectively (see Table 6-7). The impact in terms of telephony and computers is smaller but not insignificant. Table 6-7 Impact of savings on the budget and GDP Officers/Wages Telephony Computers Vehicles Budget 17.42% 0.21% 0.28% 2.80% GDP 0.271% 0.003% 0.004% 0.044% Source: WB analysis based on CENACOM (INEI, 2011-2015). Impact of efficiency in the social sphere 46. The PNP as a modern police organization does not only fight crime directly but also has other functions such as reducing the fear of crime, reducing the levels of victimization, improving the relationship between the police and the community, and others (Moore & Braga, 2003). Nonetheless, these activities are not direct police production but rather reflect the impact of police production in society. Therefore, these activities cannot be measured based on the concept of efficiency in the sense of optimization of the available resources to maximize police production. In the chain of causality, the police production is reflected in results in society, in the communities and in citizen perceptions (Moore & Braga, 2003). 47. The results of the analysis show that a more efficient PNP would have an important socioeconomic impact. Using data of the ENAPRES surveys, an estimate was made of the 182 percentages of evolution (increase or decrease) of victimization, the perception of insecurity, the confidence in the PNP, and the perception concerning performance of the PNP at the district level for the district covered in the survey. 63 The results show a positive effect in some variables. In other words, higher levels of efficiency seem to entail reductions of the victimization, reductions in the negative perception of the confidence in the PNP, and an improved perception about the PNP’s performance. The improved efficiency of the PNP does not only have a positive impact in fiscal terms and in terms of optimization of the PNP resources, but it is also reflected positively in the work of the PNP in terms of the reduction of victimization (see Figures 6-9, 6-10 and 6-11). Figure 6-9 Impact of efficiency on Figure 6-10 Impact of efficiency on victimization perceptions of the PNP Figure 6-11 Impact of police efficiency on perceptions regarding PNP performance Source: WB analysis based on CENACOM, INEI and ENAPRES (INEI, 2011-2015, 2015, 2016). Note: The horizontal axis shows the level of spending efficiency at the departmental level. The vertical axis shows the results in terms of crime and the perception regarding the PNP’s performance. Conclusions How can efficiency be improved? 48. One of the main challenges of the citizen security system is to improve the productivity of the PNP. The reallocation of inputs such as vehicles, technology, police officers and wages should go hand in hand with a process to reorganize PNP operations in the districts. This will help improve the level of security and citizen satisfaction without spending more on security. In the medium term, efficiency reforms may slow the increase of the expenditure on security. The response to the increasing crime and violence does not necessarily entail that there 63 The analysis covered a total of 477 districts. 183 should be more policemen on the streets or more vehicles and armament but rather an organizational structure to respond to the issue efficiently. 49. Considerable investments have been made in citizen security, though without asking the question “What has been the cost of that investment?” The efficiency analysis suggests substantial improvements in terms of efficiency which may simultaneously give rise to savings in terms of GDP. Table 6-8 contains a series of recommendations in terms of efficiency gains and the resulting potential savings. Likewise, the following paragraphs present and discuss the policy options to improve efficiency of the PNP. 50. The results suggest important efficiency gains. This can be achieved through the reallocation of inputs from inefficient units to more efficient units and/or by improving the organizational processes of each police station to maximize the use of resources and thus generate efficiency gains. Improvements in the planning of, for example, patrols combined with technological improvements can contribute to efficiency as well. 51. The above-mentioned elements can be supported by a resource allocation formula aimed at enhancing efficiency. The main priority of a budget allocation model of the police bodies is the distribution of the police force in areas according to the relative need of every district/department. Financing of the police bodies should not only be based on the police needs but also on the promotion of efficiency in spending in order to respond better to the present and future demands and challenges without wasting the available resources. Therefore, the formula should include an exercise to measure the technical and cost efficiency of the PNP and the efficient resource allocation to the lowest possible level of breakdown (police stations). 52. Technology can be an important factor to improve the levels of efficiency. Various studies show that the technological lag influences the levels of efficiency (de Lima and Marinho, 2017). According to the censuses of the police stations, a high percentage of the technological elements available in the Peruvian police stations are underused or are in need of repair. Moreover, the automation of PNP processes is still incipient despite the government’s efforts to expand the use of technology in police operations and processes. The use of paper for official documents still prevails in the institutions, which tends to seriously slow down the processes and decision-making. 53. Furthermore, a better coordination with the Serenazgo bodies would give rise to substantial improvements in the PNP’s efficiency levels. At present, half of the Peruvian municipalities have Serenazgo bodies, which play a fundamental role in crime prevention through joint patrols with the PNP, the protection of persons and assets, and in maintaining order in the communities. The improved coordination of activities between the PNP and the Serenazgo bodies would contribute to strengthening security in the communities, on the one hand, and on the other hand to enhancing efficiency of the PNP since the critical inputs in police operations could be used for other tasks, e.g. investigation or other crime-preventing activities. Furthermore, in the places were the Serenazgo bodies have existed for a longer time, like in the municipalities of the metropolitan region of Lima, there is a greater probability of the levels of coordination with the PNP being better than in the municipalities where the body has been implemented only recently. Accordingly, it is important for the police managers of the municipal police stations and the police 184 managers of the Serenazgo bodies to work together and plan the activities in order to enhance the efficiency and impact of the police work. 54. Finally, we cannot ignore the importance of internal organizational factors of the PNP in the efficiency of citizen security spending in Peru in the medium and long term. Even though the study has not been able to measure the effect of elements of the PNP organizational structure on the efficiency, these elements do have a significant impact on the police production. If the organizational structure is not sufficiently dynamic to adapt to the influence of factors that are internal and external to the police organization, it will continue to generate inefficiencies in the medium and long term. Therefore, the processes for organizational reforms, adjustment or improvement should keep in mind how they may influence efficiency. 185 Table 6-8 Reform Options to Enhance Efficiency of the Spending in Security Potential Future Action Efficiency Additional savings/Comments Savings (% of GDP) The potential savings could be higher if the efficiency of the police organization as a whole is examined. The results of this analysis are based only on the district police Reassignment of police officers + 0.27% of GDP stations and therefore capture less than 40 percent of the total PNP officers. Annex 4 shows the results of three illustrative examples in which there is a reallocation of PNP resources. Greater savings in GDP terms are possible because the analysis is based only on the number of vehicles in operation. There is considerable inefficiency in the use of vehicles due to the Reassignment of the vehicle fleet Minimum 0.044% of poor vehicle maintenance. Hence, the useful life of the vehicle is (cars) and/or reduction of district + GDP reduced considerably and new purchases are required to restore police stations. the vehicle fleet. In addition to this source of inefficiency, there is a need for upgrading the vehicle fleet in view of the technological advancements to combat crime and violence. The allocation of resources to citizen security is not based on a formula or a methodology. Since most of the budget finances the police organization and its operational activities, this recommendation is based on PNP financing at the district level. The main priority of a budget allocation model of the police bodies is the distribution of the police force in areas according to the Create a resource allocation relative need of every district/department. + N/A formula Even though it is difficult to include direct performance incentives without running the risk of generating perverse incentives such as, e.g. the deliberate mismanagement of criminal records, the financing should be based not only on police needs but on the promotion of spending efficiency in order to better respond to the current and future demands and challenges without ill-spending the available resources. 186 The formula should include an exercise for measuring the technical and cost efficiency of the PNP and for the efficient allocation of inputs (resources). The analysis presented in this study shows that the presence of the Serenazgo bodies lowers efficiency of the PNP. In other words, the results show that the Serenazgo is more efficient than the PNP. However, it cannot be ignored that the PNP conducts more work to control, combat and prevent crime compared to Serenazgo Improve coordination with + N/A within its limited sphere of action. Serenazgo bodies Concerning the tasks of patrolling and the prevention of crime and violence, a better activity coordination between the PNP and Serenazgo would improve efficiency of the PNP since the poorly used inputs (resources) could be reallocated to other police tasks in order to improve efficiency. 187 Introduction 1. The quantity and quality of Peru’s infrastructure is relatively low, depressing poverty reduction and growth. The density of Peru’s road network averages 0.12km per square km. Only 14 percent of the road network is paved, which is significantly less than in structural and regional peers. Moreover, the primary transport network connects Lima to other coastal cities, but not necessarily to medium and smaller cities in the Sierra or Selva, other than those important to the mining industry, limiting Peru’s economic development and reducing the competitiveness of its agricultural sector. Peru also ranks below Latin American countries (LAC), upper middle-income countries (UMIC), and OECD countries in the quality of all different types of transport infrastructure according to the World Economic Forum’s Global Competitiveness Index. Logistics costs are higher than in other countries with Peru ranking 71 out of 155 countries in the 2014 Logistics Performance Index. High logistics costs are a major constraint to poverty reduction and growth as Peru’s main exports are by and large transport intensive. They differ greatly across cities and markets, preventing the emergence of scale economies, dividing domestic markets and services, and aggravating access to foreign markets. Higher transport costs not only hamper the profitability of exporters but also of producers and service providers for the domestic market. In urban centers, underinvestment in transport infrastructure limits access to jobs and services, especially for the poor who tend to live in peripheral areas. 2. The government recognizes the importance of improving the quality of transport infrastructure, extending infrastructure in underserved regions and rapidly growing urban areas. The Strategic Sector Plan 2012-2016 of the Ministerio de Transporte y Communicaciones, for the road sector, prioritizes 22 logistics corridors which carry the bulk of Peru’s commercial and local products. Since subnational governments are responsible for improving the secondary and tertiary network, the government is pursuing a coordinated approach. Both the Plan Nacional de Logística and the Plan Nacional de Exportación are emphasized in Peru’s Agenda de Competitividad, 2014-18. The Peruvian Council on Competitiveness identified three work streams to improve transport and logistics infrastructure: i) improve the connectivity of each means of transport and at every level of government, ii) increase the supply of infrastructure through public and private investment, and iii) strengthen infrastructure management. The GOP also pursues an ambitious urban development agenda, which includes the expansion of mass transit provision. A metro network Plan for greater Lima and Callao was approved in December 2010 (DS # 059-2010- MTC) as part of a mass transit Master Plan (2010) for the Lima Metropolitan Region (LMR) which includes institutional reforms and complementary investments in the LMR and Callao. 3. This chapter analyzes spending patterns and outcomes in the transport sectors and identifies policy options to enhance efficiency. The first part analyzes the pattern of transport spending in Peru and sector outcomes relative to peer countries. It shows that reducing logistic and commuting costs will be important to sustain growth and reduce poverty given the current fiscal envelop within Peru’s medium-term fiscal consolidation. The second part seeks to identify spending inefficiencies in the areas of budget preparation and execution, investment management, and PPPs within the road and urban sector. The final section provides policy options. Public Spending and Outcomes 4. Public spending in the transport sector has increased rapidly over the last decade and started to exceed the levels of several other countries in the region . Between 2005 and 2015, public spending climbed from 0.6 percent to about 2.2 percent of GDP. In absolute terms, total expenditures increased almost nine-fold during this period. The transport sector absorbs now the third largest amount of public resources, after education and health. Public transport expenditures as a share of GDP in recent years are well above several other countries in the region (Figure ). The high share is in part explained by the surge in expenditures for a few large projects in recent years. After taking out the 5 largest transport projects, Peru’s total transport spending declines to 1.2 percent of GDP in 2014. Figure 7-1 Transport budget as a share of GDP 2.5% 2011 2013 2015 2.0% 1.5% 1.0% 0.5% 0.0% Peru Guatemala Paraguay Peru w/o Uruguay Mexico Brazil top 5 projects* Source: World Bank staff. Note: The latest data for Peru without the l5 largest transport sector are from 2014. The 5 largest transport projects in 2014 were Lima Line 2, IIRSA Sur Tramo 4: Inambari - Azángaro, IIRSA Sur Tramo 2: Urcos - Inambari, IIRSA Sur Tramo 3: Inambari - Iñapari, Eje Multimodal Amazonas Norte: Paita – Yurimaguas. 5. A steep increase in transport capital spending led to a significant expansion of Peru’s road network. Between 2005 and 2015, transport capital spending in terms of GDP increased from 0.5 percent in 2005 to 2.1 percent in 2014, before falling to 1.6 percent in 2016 (Figure 7-2). Around 70 percent of total transport spending went to the roads sector. The share of the paved core road network increased from 47 percent in 2005 to 70 percent in 2015 and the total length of the network more than doubled over the same time period. The tertiary network saw the biggest increase in length at 140 percent, followed by secondary roads with a 70 percent increase (Figure 7-3). The primary network was expanded by more than fifty percent and many primary roads were paved. Investment in urban roads increased after 2008. In 2009, around 30 percent of total transport spending went to the urban sector which declined to around 17 percent by 2015 (Figure 7-4). 189 Figure 7-2 Evolution of Transport Sector Figure 7-3 Expansion of the roads network Spending, in percent of GDP in thousands of Km. 4 200 3 150 Thousands kms 2 1 100 0 50 0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Capital expenditure (% of GDP) Current expenditure (% of GDP) Primary Secondary Tertiary Source: SIAF, MEF Source: Ministry of Transport and Communications 6. The share of transport spending by subnational government increased significantly during the last decade in line with the government’s decentralization agenda. The share of sub-national governments in transport sector expenditures climbed from 25 percent in 2005 to 61 percent in 2008, before declining to 37 percent in 2015 (Figure 7-5). Over the same period, the share of total road expenditures executed by subnational governments increased from 23 to 30 percent while the share of the Provías Decentralizado (PVD) decreased from 22 to 3 percent. The share of Provías Nacional (PVN), responsible for primary roads, increased from 52 to 67 percent. Spending decentralization was part of a broader effort to transfer responsibilities, duties, and resources from the central to subnational governments. Regional governments have been put in charge of the management of regional roads, and municipal governments of rural roads (see Box 7-1 on Transport Sector Institutions). Figure 7-4 Transport expenditure by type of Figure 7-5 Transport expenditure by level of network, share of total expenditures (%) government, Percent of GDP 100% 3.00% % of total expenditure % of GDP 80% 2.00% 60% 1.00% 40% 20% 0.00% 0% 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 National Government Regional Governments Road Rail Air Waterways Urban Other Local Governments Source: SIAF, MEF 7. Local governments spend relatively little on maintenance. Sub-national spending on maintenance has been very low (Figure 7-6). According to a BID commissioned Report 64 on the tertiary road network, the national budget allocates 26 percent of total transport expenditures to 64 Banco Interamericano de Desarrollo, 2016, “Diagnostico y propuesta de política para el mantenimiento de la red vial vecinal”. 190 current expenses, mainly through concession and performance based contracts. At the regional level, 22 percent of expenses are set aside for maintenance but only 14 percent of transport expenditures are directed to maintenance needs at the local level. A 2009 IPE Study65 estimated that only 34 percent of the total road network has received some kind of maintenance works annually and that only 12.5 percent can rely on a permanent maintenance system. 8. Maintenance of rural and regional roads not considered sufficient even to maintain recently rehabilitated roads. The inadequate attention to maintenance is reflected in road pavement conditions: only 12 percent of the tertiary network is assessed in good condition. Current expenditures on secondary roads showed sporadic improvement between 2012 and 2015 but the share of current expenditures on the tertiary network has shown no real change and the share of tertiary roads with good road conditions is low (Figure 7-7). Figure 7-6 Current versus Capital Expenditures Figure 7-7 Tertiary Network Condition: in secondary and tertiary roads By road type (2015) Dirt track Non improved path Improved path Paved 0% 50% 100% 150% % type of road Good Regular Bad No data Source: Diagnóstico y Propuesta de Política Para el Mantenimiento de la Red Vial Vecinal. Gustavo G G Picasso, Vol 1, August 4, 2016. BID. 65 Instituto Peruano de Economía (IPE), 2009, “El Reto de la Infraestructura al 2018: La Brecha de Inversión en Infraestructura en el Perú 2008”, Asociación para el Fomento de la Infraestructura Nacional, Lima, Perú. 191 Box 7-1 Transport sector institutions At the center of Peru’s transport sector is the Ministry of Transport and Communications (MTC) which is responsible for regulating, promoting, building, and supervising Peru’s transport and communications infrastructure (Figure B7-7). It includes five general directorates: (1) the General Directorate for civil aviation; (2) the General Directorate for sea and river-based transport; (3) the General Directorate for roads and railways; (4) the General Directorate for land transport; and (5) the General Directorate for social and environmental issues. The following “special programs” are also under the umbrella of the MTC: Provías Nacional and Provías Descentralizado (resulting from the merging of Provías Rural and Provías Departamental in 20061). The two Provías are public entities with private statutes which give them a certain degree of technical, administrative, and financial autonomy. Provías Nacional is in charge of managing the national road network while Provías Descentralizado provides support to subnational governments in the management and works supervision of their respective road networks. The MTC also oversees three state-owned enterprises: CORPAC1 for air transport and APN (National Port Authority) for ports and SERPOST S.A.1 for postal activities. A complex institutional system is currently in place to manage the development of public private partnerships, project execution, and regulation of the contracts signed with the private sector. In addition to MEF, there are three key actors: (1) the MTC which sets sector policies, including technical standards, and represents the GoP in awarding the concession contracts; (2) ProInversión which is in charge of promoting private sector participation and of managing the procurement process until the concession contract is awarded, and, (3) OSITRAN1, Organismo Supervisor de Inversions en Transporte, which reports to the Office of the President of the Council of Ministers, and is in charge of regulation and supervision of concession contracts. ) the Autoridad Autónoma del Sistema Electrico de Transporte Masivo de Lima y el Callao (AATE) which is the technical arm of the MTC and technical supervisor of these contracts. The Superintendencia de Transporte Terrestre de Personas Carga y Mercancías created in 2016 is an autonomous entity and the main regulator for road, freight and passenger, transport, including road safety issues, the control of vehicles and infrastructure’s compliance with the applicable standards. Figure B.7-8 Institutional framework for the Transport Sector Coordination among the multiple Peruvian institutions in charge of designing or implementing transport policies could be strengthened. The government’s objective of accelerating investments in infrastructure is challenged by institutional weaknesses at various levels. New institutions (e.g., the Superintendencia of Land Transport) have been created but it is still unclear how responsibilities will be assigned. Proper coordination needs to be defined with other areas of the MTC: DGCT1 in particular, but also CNSV on road safety or Provías Nacional on infrastructure quality monitoring. 192 Figure 7-9 Evolution of the primary network Figure 7-10 Peru Road Network 30,000 140,000 25,000 120,000 100,000 20,000 80,000 in km 15,000 in km 60,000 10,000 40,000 5,000 20,000 0 0 Primary Secondary Tertiary 1990 1995 2000 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 network network network Paved Unpaved Total Paved Unpaved Source: MTC. Source: MTC; Note: Non-Motorized Transport paths (Caminos de Herradura) are not registered 9. The size and quality of the primary road network has improved over the past decade. Between 2009 and 2015, public expenditures on the primary road network almost doubled from 3.5 to 6.2 billion Soles.66 Over the same time period, the primary network expanded and was steadily upgraded; the share of paved roads increased from 47 percent in 2005 to 70 percent in 2015 (Figure 7-9). The length of primary roads in good condition has doubled while roads in fair and poor condition have decreased by 60 percent. And 93 percent of the paved roads in the primary network are considered to be in good condition, up from 50 percent in 2010, while the largest part of secondary and tertiary roads remain unpaved (Figure 7-10). With 1.3 percent of the GDP being allocated to primary roads in 2015, as compared to 0.3 and 0.4 percent for the secondary and tertiary network, respectively, the priority afforded to the primary network is paying off with visible improvements in the quality of the network. 10. Despite these improvements, Peru lags behind peer countries with respect to the size and the quality of its road network. In fact, inadequate supply of infrastructure is considered the fourth most problematic factor for doing business in Peru (World Economic Forum 2015-2016). Only 14 percent of the country’s road network is paved, well behind Chile (29 percent), Argentina (30 percent), and Uruguay (35 percent). When compared to its population, paved road density at 745 kms per million people in Peru has improved in the last decade, but lags behind countries such as Argentina (1713 kms per million people), Chile (934 kms per million population) and Mexico (1,127 kms per million population). According to the WEF’s Global Competitiveness Report (2015-2016), Peru ranks below LAC, UMIC and OECD countries in the quality of every type of transport infrastructure but the gap is especially large in roads and railways (Figure 7-11). 66 Before 2009, roads expenditure data do not allow us to identify the amount spent by level of network (primary, secondary, tertiary). 193 Figure 7-11 Peru and the Quality of Infrastructure Index: Overall and by sub-Sector 6.0 5.3 5.4 5.1 5.1 5.0 4.5 4.2 4.3 4.1 3.8 3.8 3.9 3.6 4.0 3.6 3.5 3.7 Score (1-7) 3.2 3.0 2.9 3.0 2.1 1.9 2.0 1.0 0.0 Overall infrastructure Roads Railroad Port infrastructure Air transport infrastructure infrastructure OECD LAC UMIC Peru Source: World Economic Forum, Global Competitiveness Index 2015-2016. 11. The poor quality of roads translated into high logistics costs. Peru’s logistics costs— about 32 percent of product value—are among the highest in Latin America, well above Colombia’s (23 percent), Chile’s (18 percent), Brazil’s (26 percent), and Argentina’s (27 percent). The OECD average is only 9 percent of product value. Peru ranked 71st in the world in terms of logistic costs in 2014, substantially worse than comparator countries such as Malaysia (25), Thailand (35), and Chile (42) Briceño (2015), for instance, analyzed five export-oriented agricultural products, and found that costs related to logistics were between 20 and 50 percent of the production costs, against an average for OECD countries of 12 percent for perishable goods. The effective transport distance is by far not the only item affecting the overall cost of logistics. Cost disaggregation show that 10 to 20 percent of the total costs of logistics are related to losses during transport due to poor handling, storage, and transport and delays that lead goods to perish. Costs related to storage and switching carriers make up 20 to 30 percent of the total costs. Such high costs reflect a fragmentation of producers and transport services. For instance, the market structure for freight transport services is highly fragmented, with obsolete vehicle fleets and excessive informality of services―there were 56,504 freight service enterprises registered in the National Freight Transport Registry in 2010. Hidden costs, such as putting security measures to avoid theft, administrative costs to seek permits and licenses, and opportunity costs of having financial investments locked in due to delays, can be as high as 15 to 30 percent of the total cost of logistics. 12. As a result, Peru is lagging in accessibility, affordability, and duration of transportation. Transport infrastructure follows, to a certain extent, the population and urban footprint―accessibility is highest in the area with the largest population density, metropolitan Lima (Map 7-1, green color).67 Large mines are typically also relatively well connected to ports. But markets towards the interior, south and north of the country have low accessibility (Map 7-1, red color). These areas, however, have a comparative advantage in promising export sectors, such as agribusiness and tourism. And, they host several large cities so that the lack of connecting transport infrastructure increases the costs for service delivery and segments domestic markets, undermining competition and efficiency gains from economies of scale. Many larger cities face a 67 The transport connections in Map 7-1 are measured as the economic distances to strategic locations which are defined as Peru’s main ports, airports, border crossings, and cities with populations over 25,000 people. 194 connectivity cost mark-up of 50 percent relative to Lima. Cities in the Sierra face cost mark-ups of up to 65 to 70 percent relative to Lima for commercial or social exchanges. While Peru’s complex geography raises the cost of connecting infrastructure, markets are better connected in Colombia and Ecuador, which also have a complex geography but only small areas with low accessibility. Moreover, there is significant geographic disparity in the access to the nearest larger domestic market in Peru (Map 7-1). Map 7-1 Peru is falling behind in accessibility, affordability, and duration of transport Colombia Ecuador Peru Source: Briceño-Garmendia, C.; H. Moroz y J. Rozenberg (2015). Note: Accessibility of transport infrastructure in Peru relative to Colombia and Ecuador. The greener the area the more connected it is to ports, other cities, and trade outlets. Connectivity of a country is defined as the easiness to mobilize goods, people, and services among strategic social, service, and economic poles. Strategic locations are defined as Peru’s main ports, airports, border crossings, and cities with populations over 25,000 people. The connectivity index is measured as the amount of vehicles or goods that a dollar-equivalent of transport can mobilize over a road kilometer. Efficiency of Spending Budget planning, modification and execution by government levels 13. Budget Planning in Peru is guided by its Transport Strategy, but the strategy lacks specificity and a sector-wide transport master plan has not been prepared. The Strategic Sector Plan 2012-2016 of the Ministerio de Transporte y Communicaciones, for the road sector, prioritizes 22 logistics corridors which carry the bulk of Peru’s commercial and local products but is less clear about the strategic priorities for secondary and tertiary roads. For the road sector, the strategy could be strengthened by emphasizing the importance to: (a) complete key national corridors; (b) improving the condition of existing road assets in the secondary and tertiary networks; (c) ensuring adequate maintenance to minimize total road user costs and improved road safety; (d) involving institutions and agencies at all levels of government in identifying issues and planning road interventions; and, (e) ensuring the sustainability of the road programs by prioritizing investments in alignment with available resources while expanding the tolling network 195 to increase the revenues available. Because of the lack of a master plan, there is no systematic framework for budget allocations to both, network development and maintenance. 68 14. Consistency between original budget, modified budget, and accrued spending is higher at the central government level. Between 2005 and 2015, the modified budget of the central government has been about 14 percent higher than the approved budget (Figure 7-12). The gap between the modified and approved budgets declined in recent years and did not exceed 10 percent since 2010, apart from 2013 when transport expenditures increased significantly and the gap amounted to 13 percent. The correlation over time between the original budget and accrued spending is almost perfect (99 percent), suggesting that for primary roads the planning and budgeting process works reasonably well (Figure 7-13). Deviations from the original budget are low and execution is high for both capital spending and current spending. The correlation between the original budget and accrued spending is also high at 0.97 for capital spending and 0.95 for current spending.69 Figure 7-12 Original and modified budget Figure 7-13 Original and modified budget allocations for Road Sector, National allocations for capital spending in the Road Government Sector, National Government 8 6 7 Deflated Soles, Billions 5 6 Deflated Soles, Billions 4 5 4 3 3 2 2 1 1 - 0 2005 2006 20072008 2009 20102011 2012 20132014 2015 Accrued Modified Planned Accrued Modified Planned Source: SIAF Source: SIAF 15. For subnational governments, accrued capital spending in the road sector and the modified budget vary substantially from the originally approved budget . Contrary to the central government, regional and local governments face large upward budget revisions mid-year. Between 2005 and 2015, the difference between the original and modified budget averaged 101 percent for regional and 111 percent local government (Figure 7-14 and 7-15). In other words, the subnational governments modified budget doubled the originally approved one. The size of the budget modification has increased further in recent years. Since 2011, the budget modification 68 A master plan provides a predictable and disciplined framework for multi-year budgeting which then eases the process of year to year budgeting, keeping a country’s overall development objectives in mind. Typically, such a plan sets out institutional priorities, and also provides demand forecasts for the different modes under various economic growth scenarios, with alternative investment strategies. Typical sections in such a plan can include a situational analysis along with a gap analysis, a strategic framework, and an actual implementation plan of the objectives identified. 69 The slightly lower correlation of current spending is the result of an exceptional, but relatively large budget modification in 2012. 196 equaled a large 98 percent for regional government and increased to 126 percent for local governments. These large swings in budget allocation lead to under-execution of the budget. In fact, the correlation between accrued capital spending and the original budget execution equaled only 0.62 percent for regional governments and 0.78 for municipalities. Figure 7-14 Budgeted and Accrued Capital Figure 7-15 Budgeted and Accrued Capital Spending in the Road Sector, Regional Spending in the Road Sector, Local Government Government 2.00 2.00 1.50 1.50 Deflated Soles, Billions Deflated Soles, Billions 1.00 1.00 0.50 0.50 - - Accrued Modified Planned Accrued Modified Planned Source: SIAF Source: SIAF 16. Budget modification are exceptionally large for current spending in the roads sector at the regional level. In 2015, the modified budget for current spending was more than 5 times higher for regional government and revisions very similarly large in previous years (Figure 7-16). As a result, the correlation between the original budget and accrued expenditures is relatively low for regional governments at 0.54 percent. Budget modifications are lower, budget execution better and the correlation between executed spending and original budget at 0.89 significantly higher for local governments (Figure 7-17). 17. These differences in budget performance at different government levels translate into varying budget performance for primary, secondary and tertiary roads . Budget modifications are small for primary roads, which are the responsibility of the national government, compared to secondary and tertiary roads. Budget execution of the modified budget of the primary road network is nearly 100 percent, but ranges between 60 and 80 percent for secondary and tertiary roads (Figure 7-18 and 7-19). 197 Figure 7-16 Budgeted and Accrued Current Figure 7-17 Budgeted and Accrued Spending in the Road Sector, Regional Current Spending in the Road Sector, Government Local Government 0.6 0.6 0.5 0.5 Deflated Soles, Billions Deflated Soles, Billions 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0.0 0.0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Accrued Modified Planned Accrued Modified Planned Source: SIAF Source: SIAF Figure 7-18 Original and modified budget Figure 7-19 Budget execution of modified allocations by type of road budget by type of road 2.50 1.50 2.00 1.00 1.50 1.00 0.50 0.50 - - 2009 2010 2011 2012 2013 2014 2015 2009 2010 2011 2012 2013 2014 2015 Primary Road Network Primary Road Network Secondary Road Network Secondary Road Network Tertiary Road Network Tertiary Road Network Source: BOOST Source: BOOST 18. The large differences between the modified and approved budgets at the subnational government levels, and the resulting low subnational execution rates, are to a large extent explained by inefficiencies in the central government’s budget processes . These differences are similarly large in the water and other sectors. They are related to inefficiencies and the lack of multiyear-planning and sector spending ceilings in the central government’s budget processes. In particular, these large disparities result from a piecemeal budget allocation process which encourages line ministries to compete for scare overall resources, demanding allocations beyond their spending capacity at the beginning of the year. As a result, line ministries tend to transfer more resources to subnational entities in the second half of the year, leading to high seasonality in spending and low budget execution rates among local governments at the end of the year, both encouraging inefficient spending (see Chapter 2). 198 19. Budget execution can further be improved by better coordination between the different levels of governments and capacity building of sub-national governments . Many of the responsibilities delegated to subnational governments during the decentralization reform cannot be efficiently exercised without better coordination among the various levels of governments. The complexity of the institutional arrangements, for example, has a bearing on the budget approval process and impacts the predictability of the available maintenance funding. According to Provías Decentralizado, streamlining the budget preparation and approval process could be an important source to improve the efficiency of spending in the rural roads sector. Despite the progress seen for the primary road network in the last decade, there remain shortcomings in planning, road asset management, implementation, and budget execution capacities at the MOPC as well as with the sub-national governments, and need to be addressed urgently. 20. The unpredictability of budget approvals creates significant inefficiencies in public transport spending. The large budget revisions provide incentives to divide projects into smaller entities and to prioritize shovel-ready over higher priority investments. And for larger, multi- annual projects the unpredictability of funding can lead to implementation delays and higher costs—both in terms of capital costs since the cost of construction increases because of these delays and in terms of recurrent costs since the supervision and implementation support budgets that make up a significant part of the total recurrent costs also increase. Budget unpredictability also weakens the project prioritization process of projects since it reduces the time to implement projects within a given year as budget modifications are only approved mid-year, creating an incentive to allocate money disproportionally to shovel-ready, potentially low impact projects. Finally, the system of transfers under the decentralized system can lead to a misallocation of capital investment in transport at the local level since municipalities with more abundant resources from the Canon transfers may not have the highest investment needs. Budget planning, modification and execution of urban transport spending 21. Lima and Callao received 44 percent of total expenditures in urban transport which is consistent with the share of the urban population living in the metropolitan Lima . Both cities received more than half of all current expenditure and about 20 percent of capital expenditures. Between 2009 and 2011, Lima and Callao consumed 56 percent of total urban transport expenditures (Table 7-1). The share of capital expenditures declined significantly in 2013 (by about 20 percent) while the share in current expenditures rose. Table 7-1 Share of Urban Transport Expenditures: the top 5 cities, 2005-15 (in %) 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Lima 20.4 20.7 40.5 59.8 50.6 55.0 53.7 23.7 20.8 21.7 27.1 Callao 11.2 7.8 12.2 6.4 2.7 3.4 3.9 11.0 9.5 8.9 8.2 Cusco 0.4 1.6 2.7 1.8 6.1 4.7 5.6 8.5 7.7 5.2 7.2 Arequipa 15.6 1.9 3.1 2.2 4.0 5.1 3.7 7.0 7.0 7.2 6.4 Piura 0.0 2.8 0.1 0.6 3.4 2.7 2.9 5.6 6.6 6.7 5.8 Sum other 4 27.1 14.2 18.1 10.9 16.2 15.8 16.2 32.0 30.7 28.0 27.7 Source: BOOST data (May 2016). 199 22. There are significant differences between the accrued and planned capital expenditures across urban centers. The five largest urban areas, Lima, Callao, Cusco, Arequipa and Piura, received on average 57 percent of total urban transport expenditures between 2005 and 2015. Their actual expenditures have been twice as large as the funds allocated in the originally approved budgets, which is consistent with the fact that the modified budgets of local governments more than doubled the originally approved budgets during in that period. Regional governments are better in matching their numbers overall, with mainly Callao and Cusco significantly above budget (Figure 7-20) over time. In contrast, almost all local governments are overspending on urban transport as compared to their initial budget allocations; the ratio between the actual spending (the modified budget) and the originally approve budget is the highest in Lima (Figure 7-21). Figure 7-20 Accrued and Planned Capital Figure 7-21 Accrued and Planned Capital Expenditures for Urban Transport, Regional Expenditures for Urban Transport, Local Governments Governments Source: BOOST Source: BOOST 200 Box 7-2 Recommendations for Urban Transport Planning, Mobility, and Sustainability In Peru, there is a need for policy actions and the prioritization of investments in sustainable urban mobility, both in the mega cities (Lima and Callao) and in the intermediate or emerging cities, through which it will be possible to provide them with sustainable, efficient, safe, accessible urban mobility systems linked to urban development and land-use planning. The inevitable increase of private motorization, linked to growth of the urban footprint and of the population’s per capita income, affects the life and competitiveness of the cities. It is therefore advisable to address this issue by following three primary recommendations for urban mobility systems: Improve the integration of plans and strategies for prioritizing investments. The sector should give priority to improving the prioritization of public investments as planned, following principles of progressivity and equity in public investments, and within the fiscal frameworks at the regional and local levels. The plans and special strategic projects should be the result of the definition of priorities and the allocation of resources in line with the vision and strategic development objectives of the sector; moreover, the impact of application should be monitored and evaluated constantly in order to provide feedback for the strategy and public policy of the sector. Define the policies to enable subnational governments to improve the provision of public passenger transportation services and safer, non-motorized mobility in the cities. Several countries worldwide have adopted policy outlines for the sector within the framework of a National Urban Transport Program, in most cases with technical and financial assistance schemes for the cities. These programs usually define eligibility criteria linked to the local governments’ planning and needs so they could improve, promote and develop sustainable and multimodal urban mobility systems within the action areas of Avoid-Shift-Improve and particularly focusing on non-motorized transportation (cyclists, pedestrians and people with reduced mobility). As the cities grow and require high-capacity mobility solutions in strategic corridors (mass transport axes), the costs of investing in infrastructure and equipment are usually very high. Often subnational governments do not have either the funds or the financing capacity to cover the cost of the public works associated to the projects (CAPEX) and so national government intervention is justified in order to achieve common development objectives (at the local, national and global levels) and encourage private sector participation under sustainable operational models (OPEX). Furthermore, for non-motorized mobility solutions and demand management strategies, as measures with a lower cost and shorter deployment times, the central government’s support may be fundamental in support of these non-traditional measures that may meet with a certain degree of resistance from the population due to the lack of knowledge and the limited communication on the impact. The incentives of technical assistance and central government financing would be focused on the local governments planning and prioritizing their public investments in the most cost-effective actions to improve the population’s access to the opportunities offered by the city and reduce traffic jams, traffic-related accidents and local and global polluting emissions. The technical assistance within the framework of a national program would focus on the planning and pre-investment of prioritized actions, as well as institutional capacity-building at the local and national levels. Articulate and strengthen institutions: In conurbation zones or metropolitan areas where the institutional framework is more fragmented, international best practices suggest creating Metropolitan Transport Authorities. These authorities seek to ensure multimodal integration and complementarity of the system from a land-use planning perspective. In any case, and considering that most public transport service provision schemes involve private actors, the recommendation is to strengthen the technical and administrative capacities of the institutions to improve planning, control and oversight of the service. 201 23. Improving coordination among municipalities and with central government bodies would increase spending efficiencies in the urban transport sector . In Lima and Callao alone, there were until recently ten entities that were jointly responsible for transport expenditures. The Lima municipality, for example, is responsible for developing the bus system while the Ministry of Transport is responsible for the metro system; this situation led to investment redundancies and inefficiencies in the past. The situation is very different from Bogota, for example, where the municipality of Bogota has well defined governance structures and full autonomy to develop its urban transport infrastructure. Recently, however, the municipalities of Lima and Callao agreed to create a technical, joint transport authority, facilitating the coordination among both municipalities and reducing the political influence on investment decisions. This improvement of the governance of the local transport sector is expected to reduce bureaucratic and political costs and can have a significant impact given the very high traffic costs and air pollution in the Lima-Callao area. Construction and maintenance cost 24. Though maintenance spending tends to have a high economic return, maintenance budgets are often too small, especially for sub-national governments. Data on the costs of construction and maintenance per kilometers of road is only available at the aggregate level. Overall maintenance costs are about US$4,000 per kilometer, while construction costs range from US$150,000 to US$290,000. Road maintenance is particularly important to ensure the sustainability of existing assets.70 Sub-national governments have to comply with the technical standards set by the MTC (e.g., applicable to low-traffic roads) and by MEF (e.g., use of specific intergovernmental transfer for routine road maintenance). However, staff responsible for budget preparation are not always familiar with these standards and there may also be a need to revisit some of the parameters applied for estimating the costs of periodic and routine maintenance of rural roads. Peru has started to ensure that a significant share of its road network is adequately maintained based on relying on road concessions and new initiatives such as the microenterprises for routine maintenance program or the CREMA contracts pilot. The use of these programs should be increased and more budget resources should be allocated for maintenance for subnational roads. At the national level, the maintenance budget needs to be protected from fiscal shocks that may arise from global financial conditions or the impact of major transport investments (e.g. the second metro line). 25. Introducing a road asset management system would significantly improve the planning of maintenance spending for subnational roads . It will be important to implement maintenance program on all roads, even if initially the cost estimates are based on rules of thumb. Once a program is in place, maintenance needs can be assessed more accurately using (i) direct assessment based on the application of standard road asset management systems, or, (ii) indirect assessment based on formulas related to traffic, road length, and other variables affecting maintenance needs. This approach requires less technical capacity and has been used by Provincial Road Institutes in Peru. Moving forward, increasing the application of standard road asset management systems and building capacity at the local/ regional level would go a long way in improving the efficiency of the sector budgets. 70 Road maintenance is also the most labor-intensive of all infrastructure investments. A 2009 World Bank study estimated that, albeit limited in scope, routine maintenance works can generate up to 500 jobs per US$1 million invested and can act as a safety net mechanism for the rural poor, through the use of specialized microenterprises. 202 Table 7-2 Overall per Kilometer Maintenance and Construction Cost Estimate 2012 2013 2014 2015 Overall Kilometers 140,672 156,792 165,467 165,372 Change in Kilometers 16,120 8,674 (95) Overall Current Expenditures 1,848,727,968 1,897,134,939 2,051,708,466 2,583,792,824 Overall Capital Expenditures 5,719,451,119 7,576,345,869 7,695,534,293 7,421,995,076 Est. Maintenance per KM, Soles 13,142.08 12,099.67 12,399.53 15,624.14 Est. Construction per KM, Soles 470,000.93 887,158.24 N/A Source: World Bank staff based on data from MEF and MTC. 26. Over the next five years, investments of US$ 1,609 million need to be spent each year to totally eliminate the backlog of rehabilitation and periodic maintenance works in the total road network. There are about 203,641 km, out of which 48,774 km are paved roads. The percent of roads in sustainable very good, good, or fair condition is 69 percent for primary roads (highways and main roads), 17 percent for secondary, and 26 percent for tertiary (rural roads). While roads in fair condition require periodic maintenance works, roads in poor or very poor condition require substantially more costly rehabilitation works. The current overall road network asset value is US$16,865 million or 9 percent of GDP. 71 The overall road network carries around 22,176 million vehicle-km per year of which 55 percent travel on primary roads, 5 percent on secondary, and 40 percent on tertiary. Despite their importance, 30 percent of primary and the majority (82 percent) of tertiary roads are still unpaved. 27. Totally eliminating the backlog of rehabilitation and periodic maintenance works for paved roads requires investments of US$517 million per year over the next 5 years. 72 Out of these, 27 percent should be assigned to rehabilitation works. This investment would eliminate the rehabilitation backlog and reduce the required spending for rehabilitation and maintenance works in the following 6-20 years to US$492 million per year. For a minimum maintenance scenario, under which only rehabilitation works are done when the roads reach very poor condition, US$138 million per year are needed in the first 5 years and US$378 million per year in the following 6-20 years. Finally, under a scenario which keeps the paved roads network condition for the next 20 years at about the same level as the current condition, investments of US$319 million per year are needed in the first 5 years and US$450 million per year in the following 6-20 years (Table 7-3). 71 The current asset value is measured by the replacement value of the road network, multiplying the length of each road class with the corresponding new construction unit costs. 72 This scenario minimizes the net present value of total transport costs (rehabilitation and maintenance costs plus road user costs) over the next 20 years. The more detailed analysis focuses on the data available for paved roads of which 92 percent are in sustainable conditions (very good, good, or fair). 203 Table 7-3 Comparison of annual road maintenance and rehabilitation costs Annual Costs in Years 1 to 5 (MUS$/year) Periodic Routine Budget Maintenance Maintenance Total Scenarios Rehabilitation (M (M (M (M US$/year) US$/year) US$/year) US$/year) Optimal 135 245 138 517 Keep status quo 124 58 138 319 Minimum 1 0 138 138 Annual Costs in Years 6 to 20 (MUS$/year) Periodic Routine Budget Rehabilitation Maintenance Maintenance Total (M (M (M Scenarios (M US$/year) US$/year) US$/year) US$/year) Optimal 122 232 138 492 Keep status quo 287 25 138 450 Minimum 235 0 138 373 Source: World Bank staff based on Archondo-Callao and Sethi (2017). 28. Bringing all roads to at least fair conditions would save the Peruvian economy US$890 million over the next 20. The net present value of total transport costs (rehabilitation and maintenance costs plus road user costs) over the next 20 years would decline by US$890 million over the next 20 years under the optimal investment scenario relative to maintaining the status quo of the quality of paved roads. The savings originate from the lower road user costs which decline by US$2,298 million lower relative to the road user costs under the current quality of the paved roads network. Relative to the minimum investment scenario, the net present value of total transport costs under the optimal investment scenario is US$2,248 million lower due to savings in road user costs of US$4,972 million—indicating that for every additional dollar spent on maintenance and rehabilitation, road users save around 1.8 dollars. Table 7-4 Economic Comparison of Budget Scenarios Present Value of Costs Road Increase Decrease Budget maintenance & Road Net PV of Agency Road Scenarios rehabilitation user Total Benefits* Economic Costs (C) User Costs B/C costs costs Transport NPV Loss** * (B) * Ratio (M US$) (M US$) (M US$) (M US$) (M US$) (M US$) (M US$) (#) Optimal 4,428 61,815 66,243 2,248 0 2,724 4,972 1.8 Keep status quo 3,021 64,113 67,134 1,358 890 1,317 2,674 2.0 Minimum 1,704 66,787 68,491 0 2,248 0 0 0.0 * Comparison with do minimum scenario ** Comparison with optimal scenario Source: World Bank staff based on Archondo-Callao and Sethi (2017). 204 Public Private Participation (PPP) in the transport sector 29. Peru has awarded PPP concessions worth US$20 billion since 1996. These including both government-procured transactions and private initiatives. The expansion of PPPs in Peru’s transport sector was driven by a need to (i) accelerate public infrastructure investments within the context of a massive and aggressive investment program and low capacity of the public agency (INVIAS) to deliver works on time; and (ii) increase the execution rate of the transport sector capital and operational budgets in an environment of high fiscal revenues and significant budget allocations for infrastructure development to MTC, between 2004-2014. It was hoped that payments and commitments associated with PPPs for the next 15 years or more would help address the second challenge. PPPs allowed MTC to commit a part of future budgets and insure higher budget execution due to this commitment. Currently, there are around 90 active PPP contracts, of which one-third are in the transport sector. 30. The legal framework for PPPs was strengthened in 2015. In the past, projects considered for private sector partnership have not been subjected to an economic viability analysis or a value for money review. The earlier legal basis for development and regulation off PPPs was replaced in December, 2015, with a new legislative framework as per Legislative Decree 1224 on the Framework for the Development of Private Investment through Public-Private Partnerships and Asset Projects. The new framework sets out the principles which will inform all aspects of new project development and the rules for selecting, procuring, awarding and monitoring PPP projects, whether promoted by the state or proposed by the private sector. Key principles include: i) Requiring competition and equal treatment of all participants in procurement processes; ii) Transparency, whereby all information used in decision-making under the new law is public; iii) Orientation towards achieving results, which requires authorities to work towards facilitating private investment and identifying and removing obstacles thereto; iv)Planning, whereby the state through regional and local entities must determine the priorities and order of PPP projects, and; v) Budgetary Responsibility which specifies that consistent with the fiscal responsibility principle, the total outstanding amount of firm and contingent funding obligations for PPPs assumed by non- finance public entities can at no time exceed 12 percent of GDP. 31. But the implementation of the new legal framework is lagging behind. Peru is still lacking an infrastructure plan and procedures capable of generating a pipeline of mature projects. In addition, the institutional capacity for developing projects is often low and there is some risk of political interference. A successful implementation requires also that governments invest heavily in high quality pre-construction studies and surveys, that they structure contracts carefully to achieve an appropriate distribution of risks and expected rewards, and that the mix of private- public investment and involvement be adapted to specific circumstances. 32. Defining the long-term PPP program is important for planning and fiscal risk management. The program would assess the amount of fiscal space that PPP require as well as options to optimize the use of new PPP projects or programs . The government decided to work closely with the private sector to both close the financing gap for infrastructure investments and to stimulate the economy. According to statistics from OSITRAN, in July 2015, there were 31 transport concessions with commitments of US$14 billion. An additional USD18 billion worth of investments (public works and PPPs) were initiated in 2016. The massive scale of the investments 205 involved in implementing the government’s mass transit agenda and the payment obligations of MTC under ongoing contracts have implications for the planning and budgeting of public expenditures in the transport sector with possible repercussions beyond sectoral boundaries. The staffing and capacities of these entities have in turn significant implications for the supervision and pace of implementation and thus on the execution of the transport sector budgeted. 33. Analysis of the economic impact of PPPs has been limited so far. Value for money (VfM) to compare if projects under PPP schemes provide savings in comparison to public works could potentially reveal significant savings. The first intent to implement the VfM analysis in Peru was in 2008 but has been finalized only recently. No economic analysis has been completed for some important PPP projects, such as the Interoceanic Road, sections II, III and IV) with is associated with US$ 1.5 billion in CAPEX. 34. The coordination costs among the various institutions in charge of PPPs in the transport sector are high. Clearer responsibilities need to be assigned to the various institutions in charge of transport PPPs. The various actors involved are ProInversión, OSITRAN, MEF, MTC (Concession Unit, OGPP73, Provías Nacional). There are significant overlaps and limitations in their mandates are sometimes not clear. New legislation, or at least a Memorandum of Understanding, can help to clarify the respective roles of these institutions in the PPP project cycle. Main Challenges and Recommendations 35. Improving efficiency of public spending Peru’s road sector will be important to improve the quantity and quality of its road network within the current constrained fiscal environment. Key inefficiencies are related to:  Strengthening capacity at the subnational level to better budget, prioritize and implement investment in and maintenance of rural roads is critical . Major infrastructures such as national roads play an important role in national mobility and in connecting the different regions of Peru. However, rural roads also play a fundamental role in promoting a more inclusive growth and in rebalancing Peru’s development. Rural roads investments are also relatively inexpensive; and the institutional framework needed to implement a scaled-up rural roads program is already there: provinces already have a Provincial Road Institute, and Provías Descentralizado has accumulated years of experience in providing technical assistance to municipalities in implementing their rural roads programs.  Improving the budget allocation process and project allocation. The execution of public investments in Peru is still highly complex and bureaucratic. While several of the steps are very important to protect the quality of expenditures, a deep reengineering of the investment cycle would help simplify and optimize execution. Institutional capacity also needs to be strengthened: an institution like Provías Nacional has seen a substantial increase of its investment budget but staff and current expenditures have remained the same. While greater productivity can legitimately be expected, there is a significant risk that if nothing is done, this institution may soon reach its maximum absorptive capacity. 73 General Unit for Planning and Budget 206  Strengthening mechanisms to improve the quality of infrastructure investments, especially on issues related to public finance and safeguards. Peru’s ambitious urban transport program has emphasized acceleration of concession activities and execution of road works. While such acceleration is desirable, it should not lead to a weakening over existing mechanisms whose role is to ensure the quality of public investments. Environmental and social safeguards of such investments also need to be properly assessed. Finally, sound planning can be an effective way to prioritize investments with the highest impact. The government should strengthen these mechanisms that help improve the quality of infrastructure expenditures (planning, design studies, contingent liability evaluation, safeguards assessment, SNIP, etc.) and make them more efficient.  Refine technical norms and programming to better align maintenance with the life cycle of roads. In order to further guide its policy decisions on CREMA contracts as well as other types of maintenance contracts, Provías Nacional should initiate a study—learning from the international experience—on possible performance indicators that could help the development of maintenance contracting by level of services. The programming of maintenance activity is also critical to the optimization of road asset management. In addition to establishing a permanent and efficient routine maintenance system, it is critical that periodic maintenance be also performed on a timely basis. Modeling tools for the life cycle of roads, such as HDM-4, 74 as well as the establishment of a sound road maintenance monitoring and evaluation and a more precise definition of maintenance standards based on the different categories of roads could greatly help MTC optimize its road maintenance activities.  Fill regulatory gaps within a comprehensive strategy for transport regulation. Many regulatory instruments need to be revisited, starting with the RENAT. A comprehensive regulatory review could also be usefully initiated by the MTC. 74 Highway Design and Maintenance Standards Model, suited for the economic evaluation of road works. 207 Afonso, A., & Fernandes, S. (2008). Assessing and explaining the relative efficiency of local government. The Journal of Socio-Economics, 37(5), 1946-1979. Alda, Erik (2013). How are Police Doing in Combating Crime?: An Efficiency Analysis of the National Police in Honduras. Research. Poverty Reduction and Economic Management. The World Bank. Washington, D.C. Alda, Erik (2016). Mix and Match. Efficient Reallocation of Police Resources Through DEA and Multiobjective Optimization Modeling. Research. Justice, Law and Criminology. American University. Washington, DC. (Working Paper). Alda, Erik (2016a). Reallocation of Police Inputs to Maximize Organizational Efficiency: Evidence from a Developing Country. Research. Justice, Law and Criminology. American University. Alda, Erik (2016) “Mejorar el Desempeño en Seguridad Ciudadana a través de la Eficiencia en el Gasto”, The World Bank. Ascarza, Illich (2015). ¿Podemos mejorar la calidad del gasto en salud? [Presentación en PPT]. Lima: Oficina General de Planeamiento y Presupuesto – MINSA. Taller Banco Mundial, mayo. Barros, C. P. (2007). The city and the police force: analyzing relative efficiency in city police precincts with data envelopment analysis. International Journal of Police Science & Management, 9(2), 164-182. Becker, S. O., & Caliendo, M. (2007). Mhbounds-sensitivity analysis for average treatment effects. Bia, M., & Mattei, A. (2007). Application of the Generalized Propensity Score. Evaluation of public contributions to Piedmont enterprises. POLIS Working Paper, 80. Banco Mundial (2016). Financiamiento de la Salud en el Perú: Análisis de la Situación Actual y Opciones de Política al 2021. Washington, DC. Borges de Oliveira, Alexandre (2016).“Improving Strategic Public Procurement to leverage Fiscal Savings in Peru”, The World Bank. Cotlear, Daniel y Christel Vermeersch (2016). “Peruvian lessons for the transition from MDGs to SDGs.” The Lancet Global Health, Volume 4, Issue 6, e353-e354. Dammert, L., Ruz, F., & Salazar, F. (2008). ¿Políticas de seguridad a ciegas?: desafíos para la construcción de sistemas de información en América Latina : FLACSO-CHILE. 208 Díez-Ticio, A., & Mancebón, M. J. (2002). The efficiency of the Spanish police service: an application of the multiactivity DEA model. Applied Economics, 34(3), 351-362. DiPrete, T. A., & Gangl, M. (2004). Assessing bias in the estimation of causal effects: Rosenbaum bounds on matching estimators and instrumental variables estimation with imperfect instruments. Fulton, L. V., Bastian, N., & Shah, V. (2013). A non-linear, multi-objective, auto-optimization model for allocating resources to centrally funded hospitals. Paper presented at the Proceedings of the 12th International Decision Sciences Institute Conference. Gómez, Hernando y Roberto Steiner (2015). La reforma tributaria y su impacto sobre la tasa efectiva de tributación de las firmas en Colombia. Coyuntura Económica: Investigación Económica y Social, Volumen XLV, No 1, Junio de 2015. Gorman, M. F., & Ruggiero, J. (2008). Evaluating US state police performance using data envelopment analysis. International Journal of Production Economics, 113 (2), 1031-1037. Hirano, K., & Imbens, G. W. (2004). The propensity score with continuous treatments. Applied Bayesian modeling and causal inference from incomplete-data perspectives, 226164 , 73- 84. INEI. (2011-2015). Censo Nacional de Comisarias. SPSS. Retrieved from: http://iinei.inei.gob.pe/microdatos/ INEI. (2015). Encuesta Nacional de Programas Estratégicos -ENAPRES . INEI. (2016). INEI. INEI (2016). Perú: Encuesta Demográfica y de Salud Familiar 2015. Reporte principal . Disponible en línea. https://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1356/inde x.html INEI (2017). Perú: Evolución de los Indicadores de los objetivos de Desarrollo del Milenio 2015. Lima, Peru. Disponible en línea. http://www.inei.gob.pe/media/MenuRecursivo/publicaciones_digitales/Est/Lib1413/libro. pdf Ji, Y., & Lee, C. (2010). Data Envelopment Analysis. The Stata Journal, 10(2), 267-280. Latinobarometro. (2013). Latinobarómetro. In Latinobarómetro (Ed.). Lo, F.-Y., Chien, C.-F., & Lin, J. T. (2001). A DEA study to evaluate the relative efficiency and investigate the district reorganization of the Taiwan power company. Power Systems, IEEE Transactions on, 16(1), 170-178. 209 Londoño, J. L., & Guerrero, R. (2000). Violencia en América Latina: epidemiología y costos. Asalto al desarrollo. Violencia en América Latina , 11-57. Ministerio de Economía y FInanzas. (2013). Evaluación de Diseño y Ejecución Presupuestal- EDP. Seguridad Ciudadana-Policía Nacional del Perú . Retrieved from Lima, Perú: https://www.mef.gob.pe/contenidos/presu_publ/ppr/eval_indep/2013_resumen_seguridad .pdf MINSA 2010 (2012). Análisis de la Situación de Salud. Moore, M. H., & Braga, A. (2003). The “Bottom Line” of policing: what citizens should value (and measure!) in police performance. Paper presented at the Washington, DC: Police Executive Research Forum http://www. policeforum. org/library/policeevaluation/BottomLineofPolicing. pdf. Morante Guerrero, L. E. (2015). Modernización de la coordinación institucional entre la Policía Nacional y el serenazgo municipal. Retrieved from Lima, Perú: http://aempresarial.com/servicios/revista/86_45_QZFFNZWDHPPTUDWVZYBIKCCS GLDQRVSIUXNIXJMFZBZTEVPEUR.pdf Murray, T. (2000). Police and the challenge of the 21st century: Managing Change in Police Organisations. Canadian Police College, Ontario. Naciones Unidas (2016). Metas de Desarrollo Sostenible. http://www.un/org/sustainabledevelopment/health/. Consulta 3 de junio de 2016. Ñopo Aguilar, Hugo Rolando and Kudo, Inés (2016). “Revisión del gasto público en educación en el Perú”, The World Bank. OCDE (2014). http://www.OCDE.org/latin-america/countries/peru/. Consulta 29 de mayo 2016. Oliva, Carlos (2016) “Evolución y Desafío Fiscales en el Perú” MIMEO. OPS/OMS (2016). Análisis de los precios de los medicamentos antirretrovirales adquiridos por el Ministerio de Salud en el año 2015 versus los precios del Fondo Estratégico de OPS/OMS. Parra-Domínguez, J., Sánchez, I. M. G., & Domínguez, L. R. (2015). Relationship between police efficiency and crime rate: a worldwide approach. European Journal of Law and Economics, 39(1), 203-223 Raffo, Emma (2015). Estudio antropológico sobre cuidado de salud en el P.J. Huáscar, distrito de San Juan de Lurigancho (SJL). Informe final de consultoría. Lima: s.e. Rosenbaum, P. R. (1995). Observational studies: Springer. 210 Rosenbaum, P. R. (2005). Sensitivity Analysis in Observational Studies. In B. S. Everitt & D. C. Howell (Eds.), Encyclopedia of Statistics in Behavioral Science. Chichester: Wiley & Sons, Ltd. Shack, N., Dammert, L., & Chacón, F. (2013). EVALUACIÓN DE DISEÑO Y EJECUCIÓN PRESUPUESTAL DE: SEGURIDAD CIUDADANA –POLICÍA NACIONAL DEL PERÚ . Retrieved from Lima, Perú: https://www.mef.gob.pe/contenidos/presu_publ/ppr/eval_indep/2013_seguridad_ciudada na.pdf Simar, L., & Wilson, P. W. (2000). A general methodology for bootstrapping in non-parametric frontier models. Journal of applied statistics, 27(6), 779-802. Spottiswoode, C. (2000). Improving Police Performance. Retrieved from London, UK: Streveler, Dennis (2014). Health Information System Technical Assistance Mission Wrap-up Presentation. Lima, Peru. SUSALUD e INEI (2016). Resultados de la Encuesta Nacional de Satisfacción de Usuarios en Salud (ENSUSALUD). Presentación disponible en línea. http://portales.susalud.gob.pe/web/portal/280 UNODC. (2014). Global Study on Homicide 2013 . Retrieved from New York, NY: Vanderbilt University. (2015). Latin American Public Opinion Project. Retrieved October 2012 http://lapop.ccp.ucr.ac.cr/Lapop_English.html Van den Berg, Caroline (2016) “Addressing the large inefficiencies in water and sanitation to bring universal access within reach” , The World Bank. Vermeersch, Christel M. J. (2016). “Revisión del gasto público en salud en el Perú”, The World Bank. World Bank. (2010). Violence in the City W. Bank (Ed.) Understanding and Supporting Community Responses to Urban Violence (pp. 330). Retrieved from http://siteresources.worldbank.org/EXTSOCIALDEVELOPMENT/Resources/244362- 1164107274725/Violence_in_the_City.pdf World Bank. (2011). Crime and Violence in Central America: A Development Challenge . Retrieved from Washington, D.C.: http://siteresources.worldbank.org/INTLAC/Resources/FINAL_VOLUME_I_ENGLISH _CrimeAndViolence.pdf World Bank. (2016). BOOST-Open Budgets Portal. Retrieved from: http://wbi.worldbank.org/boost/boost-initiative 211 World Health Organization (2017). Moving from passive to strategic purchasing. http://www.who.int/health_financing/topics/purchasing/passive-to-strategic- purchasing/en/ 212