Report No. 38289-BD Bangladesh Strategy for Sustained Growth (In Two Volumes) Volume I: Summary Report June 26, 2007 Poverty Reduction and Economic Management Sector Unit South Asia Region Document of the World Bank CURRENCY EQUIVALENTS Taka (Tk) US $1.OO =Tk 69.0 (March 15,2007) GOVERNMENT'S FISCAL YEAR July 1-June 30 ABBREVIATIONS AND ACRONYMS ADP Annual Development Plan JSAN Joint Staff AdvisoryNote ASYCUDA Automatic System for CustomsData KDA Khulna DevelopmentAuthority ATC Agreementon Textilesand Clothing LDC Least Developed Country BBS BangladeshBureau of Statistics LF License Fee BERC BangladeshEnergy RegulatoryCommission LFPR Labor Force ParticipationRate BEPZA BangladeshExport ProcessingZones Authority LFS Labor Force Survey BPDB BangladeshPower DevelopmentBoard LGED Local Govemment EngineeringDepartment BRAC BangladeshRural AdvanceCommission LIC Low IncomeCountry CAR Capital Adequacy Ratio MFA Multi FiberArrangement CAS Country Assisting Strategy MTBF Medium-Term Budgetary Framework CDA Chittagong Development Authority MOU Memorandumof Understanding CEACR Committeeof Experts on the Applicationof NCBs Nationalized CommercialBanks Conventionsand Recommendations NGO Non-GovernmentalOrganization CIB Credit Information Bureau NHA National Health Accounts CMI Census of ManufacturingIndustries NPLs Non-PerformingLoans CPA ChittagongPort Authority NPSEB National Private SectorSurveyof Enterprisesin CPD Center for PolicyDialogue Bangladesh CPI ConsumerPrice Index NSS National Savings Schemes CRF Clean Report of Findings PCBs Private CommercialBanks DEDO Duty Exemption and Drawback Office PGCB Power Grid Corporation of Bangladesh DESA Dhaka Elecbicity SupplyAuthority PL Post Larvae DESCO Dhaka ElectricitySupplyCorporation PRS Poverty Reduction Strategy DFIs DevelopmentFinancial Institutions PSI Re-Shipment Inspection DPHE Departmentfor Public Health Engineering PTA PreferentialTrade Agreement DTI Direct Trader Input PWD Public Works Department DWMB Dock Workers' ManagementBoard QB QuantitativeRestrictions EBA EverythingBut Arms R&D Research and Development EPZ Export ProcessingZones RAJUK Rajdhani Unnayan Kartripakkha FA0 Food and AgricultureOrganization RD RegulatoryDuty FCBs Foreign CommercialBanks RDA Rural DevelopmentAcademy FCLs Full ContainerLoads REB Rural ElectrificationBoard FDI Foreign Direct Investment REER Real Effective ExchangeRate FPMU Food Policy MonitoringUnit RMG ReadymadeGarment FSRFD Fisheries SectorReviewand Future Development SAARC South Asian Associationof RegionalCooperation FTA Free TradeAgreement SAFTA South Asian Free Trade Area GDP Gross Domestic Product SAPTA South Asian PreferentialTrade Area GSP GeneralizedSystemof Preferences SD SupplementaryDuty HDC Human Development Center SEC Securitiesand ExchangeCommission HSC Higher SecondaryCertificate SIMA StatisticalInformationManagement& Analysis H W High Yielding Variety SME Small and Medium Enterprise IASNF Integrated Annual Surveyof Non-FarmActivities SPS Sanitaryand Phyto-Sanitary IBRD InternationalBank for Reconstructionand SSC Senior SecondaryCertificate Development SSG Ship-to-ShoreGantry TEU Twenty-foot ICA InvestmentClimateAssessment EquivalentUnit ICB InvestmentCompanyof Bangladesh TFP Total Factor Productivity ICDs Inland ContainerDepots UN United Nations ICESCR InternationalCovenantof Economic, Social VA Value Added and CulturalRights VAT Value Added Tax ICFTU InternationalConfederationof Free Trade Unions WASA Water and SewerageAuthority IDSH InfrastructureDevelopmentSurcharge WB World Bank IPOs Initial Public Offerings WBI World Bank Institute IPP IndependentPower Producer WDI World DevelopmentIndicators IRO IndustrialRelationsOrdinance WTO World Trade Organization IS0 InternationalStandardsOrganization Praful Patel, SARVP Country Director Xian Zhu, SACBD SectorDirector SadiqAhrned, SASPR SectorManager . Ijaz Nabi, SASPR Task Leader SandeepMahajan, SASPR TABLE OF CONTENTS Chapter 1: Firm-Level Productivity in Bangladesh Manufacturing Industries ......................................1 I. Introduction and Main Findings ................................................................................................... 1 I1. The Significance of Total Factor Productivity in Output Performance of the Manufacturing Sector............................................................................................................................................ 4 I11. TFP Trends in Manufacturing Industries ......................................................................................4 IV. Insignificant Role of Allocative Efficiency of Resources Within an Industry .............................7 V. Determinants of Firm-Level TFP .................................................................................................8 VI. Conclusions ................................................................................................................................14 Chapter 2: Accelerating Income Growth in Rural Bangladesh ...............................................................15 I. Introduction ................................................................................................................................15 I1. Agriculture Sector Performance: Trends and Issues...................................................................15 I11. The Rural Non-Farm Sector ....................................................................................................... 24 IV. Implications for a Rural Growth Strategy................................................................................... 25 Chapter 3: The Labor Market: Trends. Characteristics. and Challenges ..............................................28 I. Introduction ................................................................................................................................ 28 I1. Labor Market Trends in Bangladesh .......................................................................................... 29 I11. Employed Labor Force Characteristics ......................................................................................34 IV. Earnings from Paid Employment................................................................................................40 v. Determinants of Employment and Real Wages: An Analytical Approach................................ 41 VI. An Overview of Labor Laws and Regulations in Bangladesh.................................................... 43 VII. Challenges Ahead .......................................................................................................................45 Chapter 4: Trade Liberalization and Export Competitiveness ...............................................................46 I. Introduction ................................................................................................................................ 46 I1. Trends in Trade Policy and the Anti-Export Bias.......................................................................46 I11. Trade Liberalization in Bangladesh: Impact and Remaining Challenges..............................52 IV . Towards Growth-Supporting Trade Policy Strategy ..................................................................54 Chapter 5: Urban Development and Economic Growth in Bangladesh .................................................57 I. Urbanization and Urban Concentration in Bangladesh: A Process of Rapid Transformation....57 I1. Urban Concentration in Bangladesh: Extent. Causes. Benefits. and Costs ................................58 I11. Urban Management Structure and Financing and Delivery of Local Amenities........................64 IV. Policy Options To Improve the Contribution of Cities to Economic Growth: Improving Management of Cities and Removing Policy Distortions ..........................................................67 Chapter 6: Sub-National Income and Expenditure Dynamics in Bangladesh: Insights from the Macro-GDP and Micro-Survey Data ...................................................................................70 I. Introducing the Theme................................................................................................................70 I1. Income. Growth. and Inequality Variations Across Bangladesh ................................................71 I11. Correlates of Sub-National Expenditure Dynamics....................................................................75 IV. Discussion of Results and Implications for Policy .....................................................................78 Chapter 7: Governance and Growth: the Bangladeshi Conundrum ......................................................79 I. Introduction ................................................................................................................................ 79 I1. Does Governance Matter to Growth? .........................................................................................79 I11. Bangladesh's Mixed Record of Governance ..............................................................................81 IV. Unbundling Governance in Bangladesh: Some Strengths.......................................................... 83 V. Challenges in Core Governance .................................................................................................85 VI. Conclusion ..................................................................................................................................90 Chapter 8: Creating a More Efficient Financial System ........................................................................91 I. Introduction ................................................................................................................................91 I1. The Structure and Recent Performance of the Banking Sector in Bangladesh........................... 91 I11. Bangladesh's Financial System in International Comparison .................................................... 93 IV. Finance Development and Growth in Bangladesh: Is there a Link? ........................................ 96 v. Finance as Key to Faster Economic Development: Cross-country Evidence............................ 97 VI. Three Pillars of an Effective and Stable Financial System.........................................................99 VII. The Evolving Role of the Government in Bangladesh's Financial System.............................. 101 VIII. A Financial Sector Reform Agenda For Bangladesh................................................................105 IX. Conclusions ..............................................................................................................................107 Chapter 9: Infrastructure and Growth: a Focus on Power and Trade Facilitation ...........................108 I. Power Sector: Weak Governance. Poor Outcomes ..................................................................108 I1. Trade Facilitation and Growth..................................................................................................117 Annexes: ......................................................................................................................................................... 125 Statistical Annexes: ................................................................................................................................181 References .................................................................................................................................................. 197 List of Tables Table 2.1: Bangladesh: Agricultural Growth. 1981-2005............................................................. 15 Table 2.2: Bangladesh Agricultural Land and Labor Productivity. 1980181to 2004105................ 16 Table 2.3: Bangladesh: Area. Yield and Production of Rice and Wheat. 1981-2005 ....................18 Table 2.4: Bangladesh: Area. Yield and Production of Major Crops. 1981-2005..........................20 Table 2.5: Projection of Fishery SectorOutput (tons)................................................................... 22 Table 2.6: Bangladesh: Access to Servicesand Creditby Micro. Small and Medium Enterprises25 Table 3.1: Ratio of Female Population to Total Population........................................................... 30 Table 3.2: Unemploymentrate among labor forceby level of education and sex .........................33 Table 3.3: Employed Person by Statusof Employment. Gender and Locality (Percent)...............36 Table 3.4: Number of Workers by Statusand Sector..................................................................... 37 Table 3.5: Educational attainment of the economicallyactive labor force: 1996to 2003 ............. 37 Table 3.6: Formal and Informal Employed Labor Force by Sector................................................39 Table 4.1: Average Protection Rate in Bangladesh ........................................................................ 47 Table 4.2: Evolution of Import Restrictions 1991-2006................................................................. 48 Table 4.3: Bangladesh: Tariff Escalation and Cascading Impacts of Para-Tariffs on Total Nominal Protection.................................................................................................... 49 Table 4.4: Estimates of Anti-Export Bias Based on Effective ExchangeRates for Imports and Exports........................................................................................................................... 51 Table 4.5: Average MFN tariffs in South Asia and Other SelectDeveloping Countries...............52 Table 5.1: Population Growth of the Largest Agglomerations.....................................................59 Table 5.2: Urban Concentration in Comparator Counties(%) ......................................................59 Table 5.3: Specialization Index .....................................................................................................62 Table 5.4: Satisfactionwith Services(% of surveyed households)................................................63 Table 5.5: ProvidersResponsible for Delivering Urban Services..................................................66 Table 5.6: Reductions in Central Government Transfers.............................................................67 Table 6.1: GDP Levels and Growth by Division...........................................................................72 Table 6.2: Between and Within-Group Decomposition of Theil Index of Inequality at District-Level............................................................................................................ 75 Table 7.1: Investment Climate Survey........................................................................................... 83 Table 7.2: Cost of Starting a Businessin Bangladesh and Comparator Countries.........................83 Table 8.1: Structure of the Banking System(end-2005)................................................................91 Table 8.2: Performance Measure in the Banking Sector ...............................................................92 Table 8.3: Banking and Non-Banking Indicatorsfor Bangladesh in International Comparison....93 Table 9.1: Collection Rates ..........................................................................................................112 Table 9.2: Average Costs and Billing Rates.................................................................................112 List of Figures Figure 1.1. Manufacturing Sector's Share in GDP............................................................................1 Figure 1.2. Median Firm-LevelTFP Growth. Output and Input Growth (1999-2003).....................6 Figure 1.3. Decomposition on Industry-Level TFP...........................................................................7 Figure 1.4. Firm Size and Average Productivity are Negatively Correlated...................................11 Figure 1.5. Inverted-U Shaped Relationship Between Firm Age and Average Productivity ..........11 Figure 2.1. Bangladesh: Real Prices of Rice and Wheat, 1980-2003..............................................19 Figure 3.1. Age-Wise Demographic Trends in Bangladesh ............................................................30 Figure 3.2.a. Labor Force Trends ....................................................................................................... 31 Figure 3.2.b. Age Profile of Labor Force............................................................................................31 Figure 3.3.a. Changes in LFPR over Time ........................................................................................32 Figure 3.3.b. Age-LFPR Relationship............................................................................................... 32 Figure 3.4: Job CreationVs.New Labor Force Entrants................................................................. 33 Figure 3.5.: Distribution of Employment by Broad Sector: 2003.................................................... 34 Figure 3.6: Gender Decomposition of Labor Across Sectors......................................................... 35 Figure 3.7: Real Wage Trends in Major Sectors (1991normalized to 100)....................................40 Figure 3.8: Ratio of Urban Wage Rate to Rural Wage Rate for Major Economic Sectors..............41 Figure 5.1: Urbanization in Bangladesh .......................................................................................... 57 Figure 5.2: Market Accessibility across Urban Centers.................................................................. 60 Figure 6.1: District-Level Variations in Per Capita GDP (FY96-00) and Expenditure Growth Rates (FY92-05)............................................................................................... 73 Figure 6.2: District-LevelVariations in Per Capita GDP (FYOO) and Expenditure Level (FY05) .73 Figure 6.3: Per Capita GDP Growth Across the Old Districts.......................................................74 Figure 6.4: Initial Expenditure Level and Subsequent Expenditure Growth Over ..........................75 Figure 6.5: Initial Gender Inequality Measures and Expenditure Growth Over 1992-2000............76 Figure 6.6: Large-farm Landholding and Growth....................................................................... 76 Figure 6.7: Initial Inequality and Growth ........................................................................................ 77 Figure 7.1: Governance and Growth.............................................................................................. 80 Figure 7.2: Bangladesh governance Indicators 1998and 2004 ...................................................... 82 Figure 8.1: Financial Sector Penetration.......................................................................................... 94 Figure 8.2a: Sources of Firm Financing: Share of Working Capital ................................................. 96 Figure 8.2.b: Financing Obstacles: Access to and Cost of Financing.................................................96 Figure 8.3: Financial Development, Investment and Income in Bangladesh over Time .................97 Figure 8.4: GDP Per Capita Growth and Private Credit: Cross-country......................................... 98 Figure 8.5: The Interaction of Government Guarantees and Supervisory Approach..................... 105 Figure 9.1: Electricity Consumption in 2005 (kwh per capita)...................................................... 109 Figure 9.2: Installed Capacity and Available Supply....................................................................109 Figure 9.3 Electricity Sales Trends by customer Type ................................................................109 Figure 9.4: Percentage of Managers Ranking Power Shortages as a Major Constraint.................110 Figure 9.5: Institutional & Ownership Structure in the Power Sector ........................................111 Figure 9.6: Operational Performance of the Power Sector........................................................... 112 Figure 9.7: Predicted Demand-Supply of Natural Gas .................................................................. 114 List of Boxes Box 1.1: A Note on Firm-Level Data Used for Productivity Calculations ....................................2 Box 2.1: Role of Government in Agro-Food System and Agro-enterprise Development ...........27 Box 4.1: Trade Protection: Its Impact on Anti-Export Bias and Export Competitiveness...........50 Box 6.1: Regionally Disaggregated Data: Macro GDP vs.Micro Expenditure Data ..................71 Box 8.1: Three Different Approaches to the Role of Government in Financial Service Provision.......................................................................................................102 Box 8.2: Bank Privatization Process: Lessons from International Experience..........................106 Box 9.1: Vision Statement and Policy Statement: Formal Objective of Power Sector Reform.115 Box 9.2: GOB'SThree Year Road Map for Power SectorReform: 2006-08 .............................116 Box 9.3: ChittagongPort.........................................................................................................121 ACKNOWLEDGEMENTS This report was prepared by a team comprising: Sandeep Mahajan (task team leader and lead author), Tercan Baysan, Thorsten Beck, Paul Dorosh, Ana Margarida Fernandes, Somik Lall, Binayak Sen, and Alan Townsend (World Bank), Rushidan Islam (BIDS), K. Sarwar Lateef (consultant), and Md. Habibur Rahman (Bangladesh Bank). At all stages, the report has benefited tremendously from advice and support from Dr. Shankar Acharya (ICRlER), who was senior advisor to the team. Zahid Hussain and Mustafa Zakir Hussain contributed useful background pieces for the report. Mushgan Siddique provided invaluable research assistance support, Sanjana Zaman prepared the statistical tables, and F.M. Ziaul Ahsan provided additional analytical input. Oxana Bricha processed the report at various stages, and Mehar Akhter Khan, Joyce Mormita Das, and Mildred Gonzalves helped coordinate mission activities. A team of students from the Economics Department of Jahangrnagar University, led by Professor Abdul Bayes, did a commendablejob in undertaking a comprehensive firm-level survey for the report. Alfred Friendly assisted with the preparation of the Executive Summary. The report was prepared under the guidance of Sadiq Ahmed (Sector Director, SASPR), Ijaz Nabi (Sector Manager, SASPR), and Christine Wallich (previous Country Director, Bangladesh), and Xiang Zhu (Country Director, Bangladesh). The peer reviewers were Jim Hanson (formerly World Bank), Professor Siddiq Osmani (University of Ulster), and Manuela Ferro (World Bank). The team is deeply appreciative of the valuable guidance from Dr. Quazi Mesbahuddin Ahmed (former Member, General Economics Division, Ministry of Planning), the main government counterpart for the report, and Dr. Shahabuddin M. Hossain (former Advisor to the Governor, Bangladesh Bank). The team also gratefully acknowledges the useful feedback from senior government officials of General Economics Division (GED) and IMED, Ministry of Planning; Finance Division and Economic Relations Division, Ministry of Finance; Ministry of Industry; Ministry of Education, and; Bangladesh Bank. Mr. Allah Malik Kazemi, Deputy Governor, Bangladesh Bank, and Mr. Jafar Ahmed Chowdhury, Acting Member, GED, and their respective teams provided additional guidance during discussions of the "Green Cover" version of the report. A number of individuals provided very useful comments at various stages. These include Ahmad Ahsan, Syed M. Ahsan (Resident Advisor, Bangladesh Bank), Maitreyi Das, Elena Glinskaya, Prof. Wahiduddin Mahmud (Dhaka University), Deepak Mishra, Tom Rumbaugh and others in the IMF's Bangladesh team, Zaidi Sattar, Vinaya Swaroop, Hassan Zaman, and participants in workshops held in the World Bank's Dhaka office in December 2005 and October 2006. Chapter 1: Firm-Level Productivity in BangladeshManufacturing Industries 1. Introduction and Main Findings 1.1 The manufacturing sector has been a major driver of GDP growth in Bangladesh. The sector recorded an impressive 7 percent average annual growth over the 1991-2005 period, increasing its share in GDP from 13 percent to 16 percent, compared with just 5 percent average growth in the 1980s. Manufacturing exports, garment exports especially, have been the major part of the story: these now account for over 90 percent of total exports, up from 67 percent in 1981 and 80 percent in 1991. Ready- made garment (RMG) exports, increased from an insignificant amount in early 1980sto US$2.5 billion (6 percent of GDP) in FY96, and further to US$7.9 billion (13 percent of GDP) in FY06. A number of other products, such as ceramics,pharmaceuticals, and food products have also experienced strong growth over the last decade, although from a much smallerbase than garments. 1.2 Continued dynamism in the manufacturing sector would be important for Bangladesh's transition to middle-income status. Given the still modest share of manufacturing in GDP, there is considerable scope for the sector to lead the growth process for quite some time. A comparison with the East Asian economiesthat successfullymade the transition from low- to middle- income status highlights the scope for ths increase (Figure 1.1). Figure 1.1: ManufacturingSector's Share in GDP (percent) 1.3 Significant challenges would need to be overcomefor the manufacturing sector to fulfill its potential. For one, the mainstay of manufacturing activities, the garments sector (which accounts for about 40 percent of manufacturing), is facing growing competition in global markets due to the dismantling of the quota restrictions that were in place, first under the Multi-Fiber Agreement (MFA) system and then under the Agreement on Textiles and Clothing (ATC) system.' Furthermore, special arrangements for the garment sector - in the form of bonded warehouse and EPZ facilities - may have helped the sector escape some of the businesses environmentweaknesses that are endemic in Bangladesh, but, at the same time, this has also led to excessive concentration of manufacturing activities in garments, which is a source of vulnerability for the economy (see Chapter 4). To unleash the hll potential of the 1Competition for garment exports is expected to intensify even more after 2008, when the remaining shackles on Chinese garment exports to the US are removed. manufacturing sector and to achieve greater diversification, it would critical that the competitiveness of the Bangladeshimanufacturing sector be strengthenedconsiderably. Box 1.1: A Note on Firm-Level Data Used for ProductivityCalculations The analysis in this chapter uses data fiom a fm-level survey conducted for this report between November 2004 and September 2005. The survey was conducted by students of the Jahangirnagar University under the supervision of Prof. Abdul Bayes also of the same University. The survey covered a total of 682 firms in five industries: pharmaceuticals,food, ready-made garments, leatherlfootwear,and textiles. After the elimination of firms with data problems, the sample was cut to 575 firms, each fm having at most 5 years ofproduction data - fiom 1999till 2003. Annex 1provides a detailed descriptionof the data problems encountered. The majority of firms in the sample belong to the garments industry reflecting the importance of the sub-sector in manufacturing in Bangladesh. Within garments, 86% of the surveyed firms belong to the woven sub-sector, 21% to the knitwear sub-sector, and 13% to the sweater sub-sector. Most firms have more than 50 workers, althoughthe size distributionvaries significantlyacross industries. The sample comprisesrelativelyyoung firms-more than half the finns were less than 10years old while a third were 10 to 20 years old. About one-half of the firms were located in Dhaka while 17percent were located in Chittagong (excluding the EPZs). However, the location of f m differs across industries: e.g., firms in the leatherlfootwear and pharmaceuticals industries are mostly concentratedin Dhaka. The table below summarizes the main characteristicsof the surveyed firmsbased on the data for 2003. The survey produced a wealth of information on production variables, fm characteristics,and aspects of the business environment-these are summarizedin Annex Table 1. Size Distribution(% of Firms) Numberof lndustry Firms Small Medium RelativelyLarge Very Large Extremely Large ( 40 workers) (10-50 workers) (50-150 workers) (150-500 workers) (r 500workers) Pharmaceuticals 51 5.9% 15.7% 45.1% 33.3% F w d 88 1.1% 12.5% 44.3% 33.0% 9.1% Garments 276 0.4% 0.7% 48.6% 50.4% LeatherIFwtwear 24 4.2% 20.8% 33.3% 29.2% 12.5% Textiles 136 2.2% 16.9% 44.1% 36.8% Total 575 0.4% 4.0% 13.9% 44.0% 37.7% Age Distribution (% of Firms) lndustry < 5 Years 5-10 Years 10-20 Years 20-40 Years > 40 Years Old Old Old Old Old Pharmaceuticals 15.7% 113 % 29.4% 29.4% 13.7% Fwd 26.4% 10.3% 35.6% 25.3% 2.3% Garments 27.5% 27.5% 35.9% 8.3% 0.7% LeatherIFootwear 8.3% 25.0% 54.2% 12.5% Textiles 29.4% 27.2% 27.2% 11.O% 5.2% Total 25.6% 22.7% 32.8% 15.3% 3.7% Location (% of Firms) Industry Dhaka Export Chittagong Dhaka Processing Chittagong Export Khulna Other Zone ProcessingZone Pharmaceuticals 72.6% 5.9% 21.6% Food 28.4% 39.8% 4.6% 27.3% Garments 62.3% 4.7% 15.9% 9.8% 7.3% LeatherIFootwear 87.5% 4.2% 4.2% 4.2% Textiles 29.4% 1.5% 9.6% 3.7% 55.9% Total 51.3% 2.8% 16.7% 5.6% 0.7% 23.0% 1.4 This chapter seeks to deepen the understanding of thefactors that impact theproductivity and, therefore, the competitiveness of the manufacturing sector in Bangladesh. The chapter tracks the total factor productivity (TFP) performance of five major manufacturing industries over the 1999-2003 period based on a firm-level survey that was done exclusively for this report (see Box 1.1). The industries covered were ready-made garments, textiles, pharmaceuticals, food, and leathertfootwear -together these account for about 70 percent of the manufacturing sector and therefore give a good account of the sector. 1.5 Analysis shows that the average annual (median) TFP growth over 1999-2003 was positive in four out offive manufacturing industries under consideration -garments being the lone exception.2 Within the garment sector though, the average growth in the sweaters sub-sector was positive and relatively quite high. Employment generation was positive across the five industries, with annual growth rates of jobs varying between 0.7 percent in the food sector and 3 percent in the pharmaceuticals sector. Additionally, there was a sharp decline in the capital stock in all industries, likely reflecting build-up of excess capacityduring the 1990sfollowed by a period of consolidation. 1.6 Analysis also shows that allocation of resources within thepharmaceuticals, leather/footwear, and textiles industries during the sample period was inefficient: in these industries, on average, firms that were less productive received a higher share of their industry's output. However, this phenomenon is relatively insignificant in determining the overall productivity performance at the industry level: firm- level productivity is the dominant force for that. 1.7 Econometric analysis is used to determine factors that impact firm-level TFP performance, taking into account differences infirm characteristics - such as location, industry, age, and size. The results suggest that in order to strengthen TFP performance of manufacturing firms, the following areas would need particular attention: (i) The rapidly emerging energy supply constraints will need to be addressed. Results show that power supply problems are costing firms dearly in terms of TFP losses. A 1 percent increase in number of power outages in a year reduces the TFP of the average firm by 10percent. (ii) Concerted eflorts are necessaly to improve Bangladesh's attractiveness to FDI, the level of which has been quite low (less than 0.5 percent of GDP for the most part). FDI has recently picked up in extractive industries (coal and gas), telecommunications, and energy production - increasing FDI's share in GDP to about 1percent -but not in manufacturing, where the potential for productivity gains is significant. Results show that firms with any level of foreign ownership are 10percent more productive on average than firms that are wholly domestically owned. (iii) Phasing out the high anti-export bias is important. For this, it would be essential to further lower and streamline tariffs, improve trade facilitation, and tackle behind-the-border bamers to competitiveness. Results show that firms that export majority of their output are about 10percent more productive on average. This advantage to exporters may be due to technology transfers from foreign buyers and having to adopt the stringent technical and quality standards of developed markets, and from coping with tighter time and competitive pressures. As further shown by the results, the benefit of being an exporter also improves over time as valuable experience is gained in export markets. Export orientation of firms not only increases TFP levels, but also TFP growth rates. (iv) Human capital deficiencieswill need to be tackled. Results show that firm productivity improves with various measures of human capital. Higher-education levels and experience of managers in particular benefit productivity of firms, highlighting the importance of improving access to qualitytertiary education, while consolidating earlier gains on primary and secondary education. (v) Policies would need to develop the knowledge economy to strengthen the basis for more innovativeactivities and adaptation of more advanced technologies, which results show as having a positive impact on productivity of Bangladeshi manufacturing firms. Quality certifications, too, The median TFP growth rate in an industry and in any given year is the growth rate such that half the firms in that industry and year have lower TFP growth rates and the other half have higher TFP growth rates. are found to improve firm productivity, and the application of these needs to be strengthened. Such certifications guarantee the use of internationally recognized technical standards (e.g., IS0 quality certifications) and are an important means of acquiring state-of-the-art technological know-how. (vi) Thepoor law and order, often cited as major concerns by private investors in Bangladesh, will need to be improved. Results show that these hurt firm productivity: firm TFP is found to be negatively associated with protection payment by a firm relative to its sales, a proxy measure of law and order problems. (vii) Interestingly, results findjirm size and productivity to be negatively correlated - the youngest and oldest among the surveyed firms were the least productive. This suggests that Bangladeshi firms in general are unable to benefit from economies of scale. Although hard to prove at this point, it may be partly reflecting the severe corporate management deficiencies in the country. 11. The Significanceof Total Factor Productivityin Output Performance of the ManufacturingSector 1.8 What matters more for output performance -factor accumulation or TFP growth? By-and- large, there is an emerging consensus in the literature that TFP growth plays a more significant role, especially for explaining cross-country differences in GDP gr~wth.~The topic is more than just academic: it holds significant implications for policymakers - should their focus be on accumulation of capital (in other words, saving and investment) or on technology infusion via R&D, FDI, and higher- quality labor force? 1.9 What is the evidence on this from the firm-level data used in this chapter? A straightforward measure to determine the relative significance of firm-level TFP growth in output growth would be TFP growthlsales growth. The median value of this measure for each industry is presented below: 1 I Leather1 I Total Pharmaceuticals Food Garments Footwear Textiles Manufacturing TFP GrowthISales Growth 7 34 24 5 7 20 (%) of the median fm Source: Finn-leveldata collectedfor this report and staffcalculations As per this measurement, 20 percent of manufacturing sector output growth is accounted for TFP growth. But this measure does not account for output and TFP differences among firms. A more accurate assessment of the role of TFP growth in output growth would be provided by answering the following: if output growth of firm is higher than the sample average, then how much of that higher growth is attributable to TFP growth? For this, we run a regression of sales growth of each firm on TFP growth of the firm. The coefficient on TFP growth from this regression is 0.85, suggesting that 85 percent of the variation in sales growth may be attributable to TFP growth. While this figure may be on the high side, it does illustrate that TFPplays a major role in shaping output growth. This is also consistent with the findings of the cross-country studies noted above. 111. TFP Trends in ManufacturingIndustries IncreasingTFP trends over 1999-2003 1.10 This section presents trends in firm-level productivity performance of five major manufacturing industries in Bangladesh: these are readymade garments, textiles, pharmaceuticals, For example, Klenow and Rodriguez-Clare (1997) and Easterly and Levine (2000). food, and leather/footwear. Productivity here is measured by the unobservable firm-level TFP that is estimated under specific assumptions about the production function (Annex 1.2 explainsthe methodology used in obtaining the TFP measures). The firm-level survey for this exercise provides us with information on output (measuredby sales) and inputs-which include labor, material, and capital -that is needed for estimating the production function. To adjust for the effect of inflation, we deflate sales by price levels. An important advantage of our analysis relative to the literature is that we deflate sales by firm-level prices rather than the industry-level prices that are often used. 1.11 Figure 1.2 presents the median TFP trends for each of the five ind~stries.~The main results may be summarized as follows: In all industries except garments, the average annual (median) TFP growth over the 1999-2003 period was positive. Average growth in the garments sector as a whole was marginally negative over the period. With an average annual growth of 1.3 percent median TFP growth in the leather industrywas the highest. There is considerable variation within the garments sector. While the average of the annual median TFP growth in both the woven and knitwear (excluding sweaters) sub-sectors was negative over the 1999-2003period, in the sweaterssub-sectorit was positive and relativelyquite higher. Each sector and each garment sub-sector exhibited positive median TFP growth in 2002 and 2003, with the sole exception of a 0.7 percent decline in the leather industry in 2002. The pharmaceuticals and garments sectors and latitwear (excluding sweaters) and woven subsectors recovered with positive median growth in 2002 and 2003 after recording negative median growth in TFP in 2000 and 2001. The textiles sub-sector was the only sector or sub-sector with positive median growth in each of the four years under consideration. Each sector displayed positive trends in employment generation. Employment creation was particularly strong in the pharmaceuticals and garments industries, with number of employees growing annually at 2-3 percent per year. The sharp decline in the capital stock in each sector is striking. In each sector, the capital stock declined at an annual rate of at least 5 percent - the annual rate of decline was 9 percent in the case of the leather industry. This reflects relatively small investments by firms over the sample period, which is also consistent with the stagnation of the private investmentrate (in the 15.5-15.8 percent range) over 1999-2001at the national level. There has been a sharp pick-up in national private investment since 2002, which the firm-level data presented here do not reflect. It is possible that strong growth in national private investment at the rate of 13percent per annum over 1992-1999 may have led to some excess capacity, which then resulted in the subsequent slowdown. 4There are several ways to capture TFP dynamicsat the industry level after estimatingTFP at the firmlevel. One is the weighted-average TFP for firms in an industry-where the weights are assigned as per the firm size however measured. While giving the most accurate picture of the aggregate industry trends, by definition this measure is biased toward the performance of the larger firms. Alternatively measures include the unweighted-average TFP across all firms and TFP of the median fm. We prefer here the median TFP because it both gives equal weight to all firms- which seems more appealing from a policymaker's perspective - and is also more robust to presence of extreme values, unlike the simple average measure. Nonetheless, the summary TFP performance calculated by the other two methods is reported in Annex Table 3. IV. Insignificant Role of Allocative Efficiency of Resources Within an Industry 1.12 Productivity in any given industry in any given year may grow (decline)because its firms become more (less) productive andlor because output is reallocated towards the more (less) productive firms. In fact, industry level productivity (weighted-average TFP of all firms in the industry) can be broken into two components: (i) industry-level unweighted-average TFP and (ii) a term measuring the covariance between firms' shares in total sales and firms' TFP. The covariance term measures allocative efficiency: if it is positive, then the more productive firms in the industry have higher market shares and the allocation of resources is effi~ient.~ Figure 1.3: Decompositionof Industry-Level TFP Phafmacedt~calsindustry Foou industry - Unwdghted Avg. TFP Webhted Avg. TFP Unwe@ted Avg. TFP Source: ulations This positive statement has no normative content. There may be reasons why a reallocation of output to less productive firms that are, e.g., more socially or environmentally responsible, could increase economic welfare. Here, we consider only the efficient allocation of resources (output) to their more productive use. 7 1.13 As seen in Figure 1.3, the industry-levelTFP patterns are explained mostly by the firm-level TFP performance, not by the allocative efficiency of output across firms in any of the industries. 1.14 Other key features of the analysis include: In the garments industry over the entire period, and in the food industry since 2001, the allocative efficiency has made a positive contribution to industry-level productivity. In the pharmaceuticals, textiles, and leatherlfootwear industries, allocative efficiency has made a negative contribution to industry level productivity. This inefficiency in these three industries may be related to the lack of competition - including import competition - or to ineffective bankruptcy rules and the lack of markets for used capital that prevent the exit of less productive firms. V. Determinants of Firm-Level TFP 1.15 This section identifies thefactors that impact the level and growth offirm-levelproductivi~in Bangladesh. Annex 1.2 provides more details on the econometric methodology used in identifying the determinants of TFP performance.6 Specifically,the chapter considers the impact on firm-level TFP of (i) human capital, (ii) global integration, (iii) technology, (iv) finance, and (v) other factors impacting business environment. In addition, this section also assesses the role of firm size and age. Various measures are constructed in each of these areas using firm-level data survey mentioned in Box 1.1. A brief description of each factor follows. I ) Human capital here is measured by the education level and years of experience of the firm's managers, and by education-based as well as occupation-based measures of workforce skills. As shown in Annex Table 1.5, there is quite a bit of variation in the human capital characteristics within our sample of firms. In pharmaceuticals, food, and textiles industries, more than 70 percent of the f m are run by managers with post-graduate education, compared with only 54 percent in garments and 42 percent in leatherlfootwear. At the same time, the lower educational achievement of managers in the leatherlfootwear industry is somewhat compensated for by their longer work experience. Among the workers, only a small percentage is college-educated -ranging from only 3 percent in the footwearlleather industry to 24 percent in the pharmaceuticals industry. This reflects the low levels of education attainment in Bangladesh: 2.6 years on average for the population as a whole in 2000 according to the Barro and Lee (2000) statistics and 4.2 years for the working-age population as per the 2003 Labor Force Survey (Chapter 3).7 In contrast, skilled workers represent more than 60 percent of the workforce in all industries except the pharmaceuticals industry where the share is only 44 percent.8 2) Global integration occurs because of foreign ownership or by trading across borders. Annex Table 1.6 shows the corresponding summary statistics. Foreign-owned firms represent about 10 percent of the sample. The ownership composition does not vary much across industries, Existing literature suggests a list of policy variables that could potentially impact firm-levelproductivity, including export orientation, level of foreign ownership, human capital and training, and the investment climate. Many of these policy-relevant factors have also been considered in the macro literature to explaindifferences in GDP growth across countries. In testing for these policy variables in a single unified framework, the analyticalwork here, in fact, is a step over the existing literature where different strands of policy variables are generally tested for separately. Data on average educationalachievementof the workforce in the manufacturingsector is not available. A worker is classified as skilledhere if he or she falls into any of the following categories-professional, technical, administrative,managerialand skilledproduction workers. except for the leatherlfootwearindustry where a significantly smaller proportion (4 percent) is foreign-owned. Export orientation on the other hand varies significantly across industries. While almost all firms in the garments industry and 92 percent of the firms in the leatherlfootwear industry are exporters, only 26 percent of firms in the pharmaceuticals industry are. In the garments and leatherlfootwearindustries, most of the firms are majority exporters, i.e., they export over half their output. In contrast, in the pharmaceuticalsindustry there are no majority exporters. Finally, firms in the garments and footwearlleather industries have, on average, a much longer experiencein exportmarkets than firms in other ind~stries.~ 3) Technology here is measured by (i) the percentage of machinery that is computerized; (ii) the vintage of the capital stock(percentageof machinery lessthan 5 years old); (iii) whether or not the firm has any staff working on R&D activities; and (iv) whether or not the firm has been awarded one or more quality certifications (e.g., ISO)." Annex Table 1.7 shows summary statisticsfor the technology-related measures. In the garments and footwearneatherindustries, only a fraction - less than a tenth - of machinery is computerized. In contrast, in the pharmaceuticals industry it is much higher at 42 percent. While in the garments and textiles industries almost half the machinery is less than 5 years old, only 16percent of the machinery is in the leatherlfootwear industry. In the pharmaceuticals industry about 70 percent of the firms have staff engaged in some form of R&D activity,while the ratio is less than 20 percent in all other industries. Finally, the percentage of firms with one or more quality certifications varies substantially across industries: it is highest in the pharmaceuticalsindustry (40 percent) and lowest in the leatherlfootwearindustry(8 percent). 4) Several measures are used to assess the role offinancial intermediation (Annex Table 1.8). Over 60 percent of the firms report having an outstanding bank loan in each of the industries except garments where the ratio is only 30 percent. More than 60 percent of firms report having an overdraft credit line. Furthermore, between 16 percent (garments) and 49 percent (leatherlfootwear) of the firms use financing from banks and other financial intermediariesto cover any part of their working capital, and the range is between 6 percent (leatherlfootwear) and 38percent (textiles)for financing investment." 5) The business environment corresponds to the institutional, policy, and regulatory environment in which firms operate. While previous studies have mostly relied on managerial opinionsand perceptions about the business environment, our measures have the advantage of including mostly objective measurements. Bureaucratic efficiencyis measured by the number of days needed to clear customs for imports and exports and the percentage of weekly time spent by managers dealing with regulation capture. The number of power outages suffered and whether or not a firm has a generator capture the reliability of the public infrastructure. Whether or not a firm pays bribes to "get things done" and the shareof salespaid as bribes are used as measures of corruption. Finally, security payments or protection payments as a share of firm salescapturepotential crime." We use detailed information to construct the export experience variable. The f m survey asks the year since when a f m has exported, whether and in which year the f m has interrupted exports and if the f m interrupted exports, in which year exports restarted. lo Another measure of technology is the percentage of imported machinery. However, we do not use it in the regressions since it exhibits little variation across firms: 90% of firms import 100%of their machinery. Financial intermediaries here include leasing companies, special development banks, and any other public financing agency. 12 In the survey questionnaire, (i) "get things done" was explained as bribes needed for firms to resolve issues related to customs, taxes, regulations, and services, (ii) security expenses cover security guards, alarm and security systems, and (iii) protection paymentsare those to organized crime to prevent violence. Annex Table 1.9 shows summary statistics for the business environment measures. Between a fifth to a quarter of managers' time is spent in dealing with regulation and bureaucracy. Power outages appear to be a major problem for all industries, especially so in the food and leatherlfootwear industries where firms on average suffer 560 and 885 power outages, respectively,per year. In face of the frequentpower outages, it is not surprisingthat more than 80 percent of firms in all industries own a generator. There is evidence of high levels of corruption in the interactions of Bangladeshi firms with government officials and the bureaucracy. More than 85percent of the firms indicatedthat firms in their industry needed to pay bribes to government officials to get things done. Such bribes amount to 1.3 percent of sales on average in the garments industry,double of that in the other industries. Finally, firms spend minimalresources on securityexpenses and protection payments. 1.16 In addition, we also test for the impact of the size and the age offirms on theirproductivity. Studies for the US have found that firms generally enter an industry with a small size and low productivity. Firms that survive grow and converge quickly to the average size and productivity in the industry. Developed country studies have also analyzed the relationship between firm age and firm productivity. These hypotheses are yet to be tested broadly for developing countries though. Ours is the first such analysis for Bangladesh. What determines TFPlevels in Bangladesh? 1.17 This section reports thefindings of econometric analysis of the determinants offirm-level TFP for the period 1999-2003, as reported in Anna Tables 1.10 and 1.11. Annex 1.3 discusses some important econometric problems encountered in the estimations and how and to what extent these were tackled in the analysis. Firm size and productivity are inversely related. We adopt the size classification used in the Bangladesh Census of Manufacturing Industries (CMT), according which: smaflfirms have less than 10workers, medium firms have between 10and 50 workers, and largefirms have more than 50 workers. We further divide the large firm category into 3 sub-categories: relatively large firms have between 50 and 150workers, very largefirms have between 150and 500 workers, and extremely large firms have more than 500 workers. As seen in Figure 1.4, relative to the extremely large-sized firms, firms of smaller sizes are more productive.13Medium-sized firms are the most productive firms - on average 32 percent more productive than extremely large-sized firms.14Thus, contrary to findings for developed economies, in Bangladesh the larger firms are not the most productive. Although it is not possible to verify at this point, this may be because of the severe corporate management deficiencies and the resulting dearth of qualified middle managers in Bangladesh which become increasingly binding as the firm size increases, preventing exploitation of available economiesof scale.15 l3This is contrary to the predictions of theoretical models of industrial dynamics with f iheterogeneity that predict that more productive fums are larger (e.g., Jovanovic (1982)). This is confiied by empirical studies for developed economies. l4These findings are robust across different specifications. These findings also hold true if we use capital stock as the measure of fm sue: i.e., fums with larger capital stocks have lower TFP. lSOur sample is skewed towards larger firms as it includes only a few small firms (those with less than 10 workers). The small firms included are likely to be particularly efficient since they have survived and are part of the same business associations as the "bigger players" in their industry. Thus, the focus of our findings on sue and TFP Ire 1.4: Firm Size and Average Productivitv Are Negatively Correlated TFP of firms' sized 5W)+ normaftzedto 1W) Source: Firm-level data collectedfor this report and staflcalculations An inverted U-shaped pattern is seen in the relationship between firm age and productivity. The most productive firms are those that are either 10-20 years old or 20-40 years old, followed by firms that are 5-10 years old, then by firms that are more than 40 years old, and finally by firms that are less than 5 years old (Figure 1.5). This inverted U-shaped life-cycle pattern suggests that firms start at low TFP while they are young, and then improve their productivity, through learning-by-doing, engaging in innovation, and participation in internationalmarkets. Figure 1.5:Inverted-U Sha~edrelations hi^Between Firm Age and Average Productivity ,--I--1--*,-1-, 115 110 1 1 105 I j I I jII 1 g o y - 95 -- I 3r Age of Rrrn Age of finn Age of F I ~ Age or firm A m affirm I less man5 between 5 between $0 between 20 years I years ana 1 0 yean awl 20 years and 40 yews I I TFP of firms aged less than 5 years normallzed to 100 1 Source: Firm-level data collectedfor this report and staflcalculations Power supply problents lower significantly the firm TFP. Specifically, if a firm belongs to an industry and location that has 1 percent more power outages than another industry-location,then its TFP is lower by almost 10percent. The economic cost of the power sector problems that have resulted in largenation-wide deficiencies in power supply are therefore extremelyhigh. Human capital has a positive impact on firm productivity. Firms with more shlled workers (where skilled workers group includes professional, technical, administrative, managerial and slalled production workers) and firms with a higher fraction of college-educated workers are should be on the comparison across sue categories for medium-sized and particularly for large-sued f m s (i.e., those with more than 50 workers). found to be more productive, significantly so in the case of skilled workers. Firms with more experienced managers and with managers with post-graduate education have higher TFP - 3 percent higher TFP in the case of firms whose managers hold a post graduate degree. These results point out to the importance of human capital development -a dimension lagging behind in Bangladesh despite sustained improvements over the last couple of decades -for the performance of manufacturing firms. In particular, it highlights the significance of access to quality tertiary education to prepare more qualified management cadres. Global integration has apositive impact onfirm productivity. o Foreign ownershiv has a vositive imvact on firm TFP. Firms with any fraction of foreign ownership are about 10percent more productive than firms without foreign ownership.16 The usual benefits of foreign ownership - arising from both the tangible assets (e.g., better technology) and the intangible assets (e.g., better access to distribution and marketing channels and networks) -are likely to be at play here. o Exvorting has a vositive imvact on fm TFP. Firms that export the majority of their output are about 11 percent more productive on average compared with firms that don't. The TFP advantage of exporters may be due to technology transfers received from foreign buyers as well as the possibility that exporters improve their own technological capabilities in order to exploit profitable opportunities in export markets." Productivity in exporting firms also benefits from having to adopt stringent technical and quality standards of developed markets, coping with tighter time pressures, and having to face more stringent competition in general. Further, exporters face several challenges that likely result in learning-by-exporting and consequent improvements in firm TFP. To capture the learning-by-exporting effect we test for the effect of the number of years that a firm has exported. Results confirm the positive of learning-by-exportingeffect on TFP. Research and developmenthas apositive impact onfirm productivity. Firms with staff engaged in R&D activities have higher TFP. However, we do not simply interpret R&D literally as activities that bring breakthrough patentable innovations. Given the very low overall R&D levels in the country,'* it is more meaningful to ask how successful are the R&D staff in adaptation of technology and production processes to local conditions. In confirmation of this, we find that although firms with a larger percentage of new machinery (a measure of new technology) have lower TFP, that effect is counteracted when the firm has staff engaged in R&D activities. Even stronger results are obtained when the technology measure is the percentage of computerized machinery. Computerized machinery is associated with significantly higher TFP (about 16 percent higher) if the firm is engaging in R&D activity; the effect of computerized machinery on TFP is negative in itself. These findings suggest that more advanced technologies tend to improve TFPperformance but only in thepresence of significant absorptivecapacity. Similarly, quality certifications are positively and significantly associated with firm TFP. Such certifications guarantee the use of internationally recognized technical standards (e.g., IS0 quality certifications) and are an important means for firms to acquire state-of-the-art technological know-how and raise their capability to compete on global markets. l6This fmding is obtained in regressionsthat control for industry, location, and year dummy variables, therefore it is not driven by macroeconomic fluctuations (i.e., the fact that business cycles in the FDI-sending countries make some years more prone to foreign investment), nor by a composition effect (i.e., the fact that certain industries are more prone to receive FDI), and it is also not driven by a location effect (i.e., the fact that FDI f m are more likely to locate in certainregions such as EPZs). l7Westphal(2002) documentsthe latter possibility for f m in Taiwan. l8R&D expendituresrepresent only 0.03% of GDP in Bangladesh, compared to 0.7% in China and India, and 0.2% in the Philippines. Potential for crime dampens TFPperformance. Firms making larger protection payments are significantlyless productive than other firms. We assume that larger protection payments to be "spared" from organized crime are a proxy for an environment with more potential crime. In unreported regressions, we also find a negative effect of the ratio of security expenses to sales when that variable is the proxy for potential crime. No clear relationship betweenfinancial intermediation andfirm TFP. Firms with an overdraft facility or credit line have higher TFP, but this relationship is not statistically significant. In contrast, firms with access to a bank loan have significantly lower TFP. Qualitatively similar findings are obtained in the regressions that test only for the impact of using the overdraft facility or with access to a bank loan along with some basic control variable^.'^ In unreported regressions,we do find a positive correlation between the percentage of working capital financed by banks and related institutionsand firm TFP. These intriguing findings likely have to do with the inefficiencies of the banking sector in Bangladesh and deserve a more detailed analysis. However, they should by no means be taken as evidence that access to external finance is not relevant for firm performance. Some counterintuitive results emerge in looking at the linkage between TFP and corruption/red tape. For example, the percentage of management time spent on dealing with regulations is found to be positively correlated with TFP. This could easily be because better performing firms are more visible to public officials which makes them more subject to inspectionsor visits from officials. Moreover, firms with higher TFP may also be more involved in activities - such as firm expansion, exports, etc. - that require permits andlor meetings with officials. Similarly,we find that fm TFP is higher in industries and locationswhere firms pay a larger percentage of their sales in bribes to get things done. Again, even though counterintuitive, it is quite plausible that this is because firms with higher TFP can afford to pay larger bribes to cut through bureaucratichassles. Finally,the number of days needed to clear customs for imports does not seemto impact firm TFP. What determines TFPgrowth in Bangladesh? 1.18 Results of the econometric estimationsof the factors that affect the change in TFP between 1999- 2003 are presented in Annex Table 1.12. Summary results follow: There is strong evidence of conditional TFP convergence. Results show consistently and significantlythat firms with lower initial TFP level tended to grow faster. Over time, this process would lead to convergencein firms' TFP to the average TFP in the f m s ' industry and location. The size of thefirm is inversely related to TFPgrowth. Medium-sized and large-sized firms are found to have significantly higher TFP growth than extremely large-sized firms. A possible interpretation is that smaller firms may be more flexible in adapting to shocks. Or, it could be that corporate management deficienciesbecome more binding as the firm gets larger, inhibiting its growth. Since the TFP level too was found to be inversely related to firm size, it seems that the gap between smaller and larger firms in terms of their productivityis widening over time. l9 There is widespread evidence of a correlation between firm sue and access to finance. In fact, sue is often used as a proxy for financing constraints (see, e.g., Beck, Demirgiiq-Kunt, and Maksimovic (2005)). The fmding that firms with access to a bank loan have lower TFP is unchanged, however, when we estimate our regressions excluding sue but including the other basic controls. Thus, the negative effect of the loan dummy variable is not drivenby the potential correlation between firm sue and access to finance. Wefind an inverse relationship between firm age and TFP growth. Older firms have lower TFP growth than f m s that are less than five years old. Although not significant,the findingson age are similar to those obtained for firms in African countries.20 Global integration is associated with faster TFP growth. Firms that export majority of their output seem to enjoy faster TFP growth. In fact, being a majority exporter increasesTFP growth by 0.7 percentage points per year on average. Learning-by-exporting seems to be at play here as firms with more experience in exporting are also those with higher TFP growth. Foreign owned firms also seem to have higher TFP growth, although this particular result is statistically not significant. Firms with more skilled workers have significantlyfaster TFPgrowth. Also, firms with higher levels of external traininghave higher TFP growth. Degree of red tape hurts TFPgrowth. Firms whose managers spend a larger percentageof their time dealing with regulation have lower TFP growth rates. A 10 percentage point increase in percent of managers' time spent on regulation can lower TFP growth by 2.5 percentagepoints. VI. Conclusions 1.19 The findings of this chapter highlight several important areas of policy relevance in which improvements are likely to bring large benefits for firm-level productivity in Bangladesh. The importance of accelerating infrastructuredevelopment stands out starkly. In addition, policies to promote human capital formation at various levels, incentivize R&D activities by firms, and encourage use of quality certifications hold significant potential in terms of improving productivity in manufacturing activities. Deepening of global integration by both promoting export orientation among manufacturing firms and enabling foreign ownership of manufacturing activities in Bangladesh is another high pay-off reform area. 1.20 Finally, the analytical work in this chapter points to the benefits of systematically collecting information for the manufacturing sector. The Government would be well advised to institutionalizing again a comprehensiveand regular census of manufacturingfirms in Bangladesh. 20Van Biesebroeck(2005) Chapter 2: Accelerating Income Growthin Rural Bangladesh I. Introduction 2.1. Bangladesh has enjoyed major successes in rural development, in both agriculture and the non-farm sector. Agricultural output has grown at an average rate of 2.9 percent since 1980, outpacing population growth and allowing the country to strengthen its food security and make a significant dent on rural poverty. In particular,rice production -which accountsfor two-thirds of total value added of crops -hasmore thandoubledsinceindependenceasaresultofpublicandprivate investmentsin irrigation, dissemination of improved seeds and fertilizer, and a policy environment that maintained incentives for domestic production. At the same time, significant investments in roads contributed to the development of well-functioning markets for major agriculturalproducts. The rural non-farm sector - employer of 40 percent of rural workforce-has also been growing fast, at over 4 percent per year since the early 1990s. 2.2. However, the rural sector is faced with a number of emerging challenges in order to sustain good performance. Agriculture is under pressure to improve its productivity due to a shrinking land base. As urban development continues,area cultivated has declining steadily. As a result, total area sown has also declined, and, in fact, the rate of decline is expected to accelerate. Furthermore, agricultural diversification is lacking, with potential in higher-end products largely unexploited: agriculture production is dominated by crops which, in turn, are largely rice based. The rural non-farm sector is faced with a number of structural constraints that (if unaddressed) risk undermining its future growth prospects. Sharp productivity increases in both agriculture and rural non-farm sector are necessary to support sustained high growth and faster poverty reduction. This calls for the next generation of agro reforms and addressing the major investment climateconstraintsfaced by rural enterprises. II. Agriculture Sector Performance: Trends and Issues 2.3. Agriculture sector growth has been sporadic. The sector recorded relatively good growth of about 5 percent in the second half of the 1990s, recovering from weak performance in the first half. Growth has since fallen to its long-term trend of just under 3 percent in the 2000s (Table 2.1). In common with other fast growing developing countries, the rest of the economy has outpaced the agriculturesector,resulting in a decline in the sector's share in (producer price) GDP, from 30 percent in 1990to 20 percent in FY06. Table 2.1: Bangladesh: AgriculturalGrowth, 1981-2005 All Agriculture Crops Livestock Forestry Fish Growth Rates Shareof GDP at Producer Price (%, period average) 1981-90 31.5 23.1 3.5 1.6 3.2 1991-00 26.7 16.2 3.4 2.0 5.1 2001-05 21.9 12.6 2.6 1.8 4.9 1981-05 27.6 18.2 3.3 1.8 4.3 Source: Calculatedfrom BBS data. 2.4. The sector is dominated by crops, although fish production has gained considerable share since the 1980s. Crops account for about 60 percent of agricultural GDP (AGDP), compared with 73 percent in 1980. The slack was mostly picked up by fisheries, the fastest growing sub-sector in agriculture, which saw its share increasing from 10 percent to 23 percent in AGDP over the 1980-2000 period, benefiting especially from the high growth in the 1990s(Table 2.1). Fish production growth has slackened significantly since 2001, however, resulting in fisheries' share in AGDP falling to 21 percent by FY06. The livestock sub-sector (including poultry) has seen a small decrease in its share in AGDP since 1980(from 13percent of AGDP to 12percent), while forestry has doubled its share to 8percent. 2.5. Intensifying landpressures are a major constraint on agricultural growth. Owing to extremely high population density (one of the highest in the world) and rapid urbanization, the net cultivable area declined by an average rate of 2 percent per year in the first half of the 1990sand by close to 1percent in the second half, as potential agricultural land was diverted to other uses such as housing, roads, and industrial development (Table 2.2). The net area sown has also been on a decline, albeit at a more gradual pace, resulting in an increase in the ratio of net area sown to total cultivable area from 85 percent in FY90 to 95 percent in FY03. With less than 5 percent of total cultivable land not sown, further shrinkage in cultivableland would lead to an almost equivalent decline in net area sown. Table 2.2: BangladeshAgriculturalLand and Labor Productivity, 1980181 to 2004105 Net Cultivable Net Area Sown Crop VA VA (Tk bn Land 1996 prices)/Net Million Hectares Million Hectares (Tk bn 1996 prices) Area Sown (mill. Hectare) 1981-90 9.5 1991-95 9.0 1996-00 8.4 2001-06 8.6 1981-06 9.0 Average Annual Growth 1982-90 0.5 1991-95 -2.1 1996-00 -0.7 2001-06 -0.1 1982-06 -0.4 Source:BBS and Ministry ofAgriculture. Note: Data on cultivable area and area sown are available only through 2002. Data for these variables for 2003 through 2006 assume no changes relative to 2002. 2.6. Land productivity has been improving since the early 1980s (Table 2.2) due to increased productivill/hectare of individual crops, increased multiple cropping, and, to some extent, a shift to higher valued crop production.21 Given that net cultivable land and net area sown have reached their limits, future improvements in the crop sector performance will depend mainly on productivity growth. 21There is no statistical evidence that land fragmentation has significantly lowered total factor productivity in Bangladeshi crop agriculture. Econometric evidence from Pakistan suggests that plot size does not affect productivity there, controlling for crop choice and other factors (Pakistan Poverty Assessment, World Bank (2002), and; Pakistan Rural Factor Markets: Policy Reforms for Growth and Equity, World Bank (2004),). The Green Revolution agricultural technology (improved seeds, irrigation and fertilizer) is to a large extent scale-neutral (i.e. there are little or no economies of scale in increasing farm size), particularly for the major grains (rice, wheat, maize). Some scale economies may exist in marketing, particularly of highly perishable fresh fruits and vegetables. There is evidence of economies to scale in aquaculture, however, particularly in coastal areas where the need to coordinate the timing of water flows may make large farms more efficient than small farms. For one, this will require continued improvements in rice yields, especially in the aman season when the HYV rice variety is used only in half the sown area. Ultimately though, raising real agricultural incomes will require more diversification into higher-valued crops and increased output of the livestock and fishing sub-sectors. In particular, Bangladesh would need to better avail of its long-term competitive advantage in inland aquaculture, where productivity performance in recent years has fallen well short of potential. Crops Sub-sector 2.7. Sustained growth of cropsproduction -rice and wheat, inparticular -in the 1980s and 1990s enabled Bangladesh to achieve itsfood grain production targets. Growth in foodgrain production has outpaced population growth, eliminating the so-called "food gap" (calculated as the difference between the amount of foodgrain required to meet the consumption target of 454 grams of foodgrain per person per day and net domestic production) by FYOO. However, growth has slowed down considerably since FY02, averaging 1.8 percent per year over FY02-06, compared with an average 2.4 percent over FY81- 01. This was essentiallyon account of a slowdownin growth of rice production, a sharp decline in wheat production (which fell by 12percent per year, on average, over the FY02-05 period), and continued poor performances of other crops such asjute, sugarcane,pulses, and oilseeds. 2.8. Diversification of cropproduction to non-rice crops, especially the higher value added ones, is important, but is being held back by a number offactors. Marketing constraints including shortage of cold storage facilities and a functioning cold chain, lack of grades and standards, insufficient processing capacity and, to some extent, weak consumer demand related to low incomes and consumer preferences remain a major impediment. Transport bottlenecks also slow delivery of fresh vegetable products and lead to reduced qualityand salesprice. Rice and Wheat Production 2.9. Rice production, which accountsfor two-thirds of crops value-added of; doubled between 1980 and 2005, growing at an average rate of 3 percent per year. Growth has been on a declining trend, however, falling from an average 3.7 percent during the 1980sto 2.7 percent in the 1990sand further to 1.9 percent in the 2000s. Growth was also quite uneven in the 1990s;averaging -1 percent a year in the first half of the decade (due to droughts and problems with fertilizer supply and distribution) and a whopping 6.6 percent in the second half. 2.10. Growth in riceproduction has been underpinned by improvements in land productivity, as the land area under cultivation has not changed much. In fact, most of the increase in rice production since 1980has been on account of the winter season (January-May)boro crop (Table 2.3)." Traditionally, the main rice crop in Bangladesh was the monsoon season aman crop (August-December), which depends almost entirely on rainfall and regular seasonal flooding of rivers and streams. Following the liberalization of imports of diesel engines and pumps for tubewell irrigation and expansion in fertilizer in the 1980s and with growing use of high yielding varieties of rice (HW's), the area planted to boro increased sharply.23 The boro area has almost quadrupled since 1980, replacing the lower yielding aus rice crop (April-August) in many areas. In the meanwhile, as a result of concomitant productivity improvements, boro production has increased almost six-fold. The area planted to aman crop, in the meanwhile, has shrunk by about 10 percent and although its yield has improved the rate of improvement has been declining. 22The dates of planting and harvest of the aman and other rice crops given in this section are indicative; actual dates vary slightly across Bangladesh. 23Ahmed 2001. Table 2.3: Bangladesh: Area, Yield and Production of Rice and Wheat, 1981-2005 Annual Average Growth Rates 1981-90 1991-2000 2001-05 1981-90 1991-2000 2001-2005 Aus Production (th tons) 2992 1900 1781 -0.78 -3.27 -2.38 Area (th ha's) 2906 1655 1204 -2.75 -4.94 -5.23 Yield (tonsha) 1.03 1.15 1.48 1.94 1.73 2.81 Aman Production (th tons) 7920 9127 10886 3.02 1.81 -0.59 Area (th ha's) 5872 5683 5593 -0.30 0.13 -1.49 Yield (tonsha) 1.35 1.61 1.95 3.07 1.39 0.77 Boro Production (th tons) 4100 7745 12517 10.20 6.33 4.70 Area (th ha's) 1666 2863 3877 8.46 3.99 2.17 Yield (tondha) 2.46 2.71 3.22 1.58 2.16 2.46 Total Rice Production (th tons) 15011 18772 25184 3.71 2.75 1.86 Area (th ha's) 10444 10201 10682 0.33 0.24 -0.62 Yield (tonsha) 1.44 1.84 2.36 3.36 2.46 2.46 Wheat Production (th tons) 1092 1400 1403 2.26 7.85 -11.64 Area (th ha's) 572 699 684 4.36 3.68 -7.61 Yield (tonsha) 1.91 2.00 2.05 -1.88 4.04 -4.50 Total Production (th tons) 16103 20172 26587 3.56 3.00 1.07 Area (th ha's) 11016 10900 11367 0.46 0.44 -1.07 Yield (tonsha) 1.46 1.85 2.34 3.06 2.52 2.12 Source: Staff calculationsfrom BBS data. 2.11. Production of wheat (the second largest crop) has also seen a substantial increase since the early 1970s, although production levels have declined sharply after peaking in FY99 (Table 2.3). Wheat production increased from an average of about 100 thousand tons per year during FY70-75, to an average of more than 1.8 million tons in FYOO, due to a seven-fold expansion in area and a doubling of wheat yields. Production growth was especially rapid in the 1970s (37 percent per year), as area increased by 19 percent per year and yields rose by 15 percent per year, and also in the 1990s when production increased by 83 percent and area under cultivation by 39 percent. In recent years, however, area under wheat production has declined from a peak of 0.9 million hectares in FY99 to only 0.6 million hectares in FY05, as area planted to boro rice, maize and potatoes expanded. In the meanwhile, per acre yield has dropped by more than 20 percent since FY99, resulting in halving of wheat output since then. 2.12. The long-term increases in rice and wheat production have resulted in declining domestic prices (Figure 2.1.).'~ From the late 1970s to the early 1990s,real rice prices fell by about 30 percent. 24Imports of rice over these periods were rather small (about 250 thousand tons per year in the 1980s and 700 thousand tons per year in the 1990s (Dorosh, 2001). Real rice prices have, more or less, leveled off since, although with substantialfluctuations. Nonetheless, real incomes of farmers who adopted new technology, particularly those who increased their area planted with boro rice, have improved in general. Farmers unable to adopt the new technology because of lack of imgation, appropriate drainage or other constraints, particularly in southern and northeastern parts of Bangladesh, however, may have experienced declines in real incomes.25 Lower real prices would also have directly benefited consumers of rice, a staple for the majority. Moreover, the increased size of the rice harvest and the more even seasonal spread between the monsoon season (arnan) rice crop and the winter season boro crop have helped prevent large spikes in price following years of drought or floods that damaged the arnan In addition, private sector rice imports (made possible by trade liberalization in the early 1990s)have added to price stability in years of poor rice harvests (FY95, FY98 and FY99), by keeping domestic prices from rising above the cost of imports.27 Figure 2.1: Bangladesh: Real Prices of Rice and Wheat, 1980-2003 Note: Prices are deflated using the non-food Dhaka middle-income Cost of Living Index (and the national CPI after June 1998). Source: FPMUdata and author's calculation. 2.13. Despite the improvements, rice yields in Bangladesh remain substantially lower than in the East Asian economies, although better than in India and Pakistan. Wheatyields, on the other hand, are significantly lower than both the South Asian and East Asian comparators. Average rice yields in Bangladesh are about 5 to 10 percent higher than in India and Palustan, but 45,20, and 25 percent lower than in China, Vietnam and Indonesia, respectively (FA0 production data). Wheat yields in Bangladesh are about 30 percent lower than in India, 15percent lower than in Pakistan, and are less than half of those in China. Rice cultivation in Bangladesh lacks the water control (in aus and arnan seasons) and warm temperatures (in winterlboro season) enjoyed in the East Asian countries. In addition, fertilizer use is significantly lower than in China. Bangladesh, however, had higher yield growth of both rice and wheat over the 1980-2003than India and Pakistan. 25SeeAhmed (2001)and Arndt et a1(2002). 26In addition, changes in cropping patterns resulting from an increase in minor irrigation and availabilityof modem varieties have reduced the susceptibility of Bangladesh agriculture to floods by leading to a reduction in deepwater (broadcast) aman rice, grown on flood-proneland during the monsoon season; and greatly increasingthe quantity of rice produced within five to six months of any damaged monsoon season rice harvest (Hossain, Bose and Chowdhury, 2001; del Ninno, Dorosh, Smith and Roy, 2001; Chowdhuryand Haggblade, 2000). 27See Dorosh (2001). 2.14. Sustained increases in rice production through yield improvements are needed to keep real prices from rising, as well as to free up land for diversijication into high-valued crops. The comparisons with the East Asian economies suggest that significant scope for further improving rice productivity exist. Apart from aman rice, few crops can be profitably grown on flooded land during the monsoon season (jute is an exception in some areas). Since only about half of the aman crop (53 percent in FY04) is cultivated with HYV varieties, (compared with 95 percent of boro), continued expansion of HYV cultivation in the aman season is one potential source of further rice productivity and production growth. In the winter season, many crops compete with boro rice, and greater diversification of agricultural production has the potential to increase rural incomes and lead to increased consumption of non-foodgrains in the citizens' diet.28 Table 2.4: Bangladesh: Area, Yield and Production of Major Crops, 1981-2005 Annual Average Growth Rates 1981-90 1991-2000 2001-2005 1981-90 1991-2000 1991-2005 Maize Production (th tons) 2 3 307 15.01 0.79 46.10 Area (th ha's) 3 3 55 6.19 -1.04 26.64 Yield (tonsha) 0.76 1.05 5.66 8.30 1.85 15.36 Jute Production (th tons) 977 879 819 -0.06 -1.05 -0.89 Area (th ha's) 658 514 437 -0.95 -2.06 -2.12 Yield (tonsha) 1.49 1.71 1.87 0.90 1.03 1.25 Sugarcane Production (th tons) 6970 7312 6642 0.32 -0.70 -1.05 Area (th ha's) 166 179 165 1.62 -1.03 -1.07 Yield (tonstha) 41.99 40.83 40.20 -1.28 0.33 0.03 Pulses Production (th tons) 430 500 348 12.03 -2.09 -3.64 Area (th ha's) 619 666 449 11.19 -3.15 -4.36 Yield (tonstha) 0.69 0.75 0.78 0.76 1.09 0.75 Oilseeds Production (th tons) 424 462 281 6.84 -0.20 -3.91 Area (th ha's) 496 544 365 8.15 -1.68 -3.81 Yield (tonstha) 0.85 0.85 0.77 -1.21 1.50 -0.10 Potatoes Production (th tons) 1115 1715 3376 0.55 8.37 9.66 Area (th ha's) 107 153 251 0.00 6.30 6.99 Yield (tonstha) 10.59 11.11 13.44 0.55 1.95 2.50 Source: Staff'calculationsfi-om BBS data. Note: Data forjute, sugarcane,pulses, oilseeds and potatoes are through 2003104. 28Note also that there remains a substantial yield gap between national average yields and experiment station results ranging from 29 percent for boro to 53 percent for aus (Revitalizing the Agricultural Technology System in Bangladesh, World Bank, 2005). Part of this yield gap is due to improved management, which may require increased labor and other inputs that are not necessarilyfinanciallyoptimal from the farmers' perspective. Other Crops 2.15. Maize production has expanded by nearly I00 times in the past seven years (Table 2.4), with area cultivated and production increasing from only 2.8 thousand hectares and 2.6 thousand tons in FY98 to 102 thousand hectares and 230 thousand tons in FY05. Maize production has benefited from dissemination and adoption of highly profitable hybrid seeds (essentially all maize in Bangladesh is now planted with hybrids) and ample domestic demand for poultry feed. Production is concentrated in the northwest comer of Bangladesh (Thakurgaon,Panchogarh, Dinajpur, Nilphamari) and in parts of western Bangladesh (Kushtia and Jessore). 2.16. Potato production increased by 7.0 percent per year from FY90 to FY04. Area planted to potatoes surged by 120 thousand hectares (an 80 percent increase) in FY99 after the opening of the Jamuna bridge that helped connect potential growing areas around Bogra with the Dhaka market. Since then, area under cultivation has remained steady, but production has continued to increase due to yield increases;yield increased by 28 percent between FY99 and FY03. 2.17. In contrast, area andproduction ofjute, sugarcane,pulses and oilseeds have declined over the last I5 years. Area planted to jute, once a major export of Bangladesh, declined by an average of 0.9 percent per year between FY90 to FY04. Likewise, area planted to sugarcane declined by an average of 1.1 percent per year. The declines in pulses (3.6 percent per year) and oilseeds (3.9 percent per year) were even steeper,as there has been little improvement in technology in these crops, unlike in rice, wheat, maize and potatoes. Non-Crop Agriculture 2.18. Non-crop agriculture,fish and poultry in particular, has shown good growth since the early 1990s and continues to hold significant potential for growth, although there are emerging concerns that need to be addressed. The fish sub-sector,nearly 80percent of which derives from inland water (i.e. rivers, ponds, flood plains, etc.), grew at a rapid pace of 8 percent per year during the 1990s, but then experienced a major setback in 2001 with a 4.5 percent decline in output; growth has since recovered somewhat to about 3 percent. In the 1990s, fish output was buoyed by doubling of frozen food exports (mostly shnmps) during the first half of the 1990s. Shrimp exports-the second largest source of foreign exchange and employer of about 600,000 workers - fell during FY96-99 due in large measure to problems in meeting sanitary and phyto-sanitary (SPS) requirements of importers, and remained at the FY96 level throughN03. It also appears that expansion of the poultry industrymay be slowing down on account of subdued demand and high cost of feed. Fish and ~ h n m u ~ ~ 2.19. Fish production has increased rapidly since the early 1990s, mainly due to increases in aqua- culture Cfish raised in ponds); output of the capture fisheries sub-sector has stagnated. Within this basic division between aqua-culture and capture fishing, there are also substantialdifferences in the types of fish, production growth rates, and distribution of benefits between inland aquaculture (fresh water fish ponds) and brackish-water aquaculture (mainly shrimp), and between inland capture fisheries (rivers, floodplains, ox-bow lakes and the Sundarbans)and marine capture fisheries (including both trawling, i.e. Industrial coastal fishing and artisanal, i.e. small-scale fishing) (Table 2.5). The aquaculturesub-sectors 29This section is based largely on the World Bank BangladeshWater CAS (2005), supplemented with material from other major studies on fish and shrimp in Bangladesh, including Collis (2004), Revitalizing the Agricultural TechnologySystem in Bangladesh (World Bank 2005), Cato and Subasinge(2004), and Dobson and Quader (2004). (inland and brackish water) have the potential for signiJicant sustained growth. Capturefishing, by contrast, is likely to continuetostagnate. Table 2.5: Projectionsof Fishery Sector Output (tons) 2002 2012 (estimate) Annual change Production sector (mtlyear) (mffyear) (%) Inland aquaculture 850,000 1,465,744 5.6 Coastal aauaculture 94.580 129.597 3.2 Inland capture fisheries 750,419 606,919 - 2.1 Coastallmarinecapture fisheries 589,500 501,689 - 1.6 Total 2,284,499 2,703,949 1.5 Source: Fisheries Sector Review and Future Development (FSRFD); WorldBank Bangladesh Water Country - Assistance Strategy 2005. Aquaculture Fisheries 2.20. Production of both inland and brackish-water aquaculture has grown rapidly since the early 1990s - at over 10percent per year. Most of the growth has derived from utilization of existing ponds and established technologies. Relatively high growth rates are expected as long as internal prices remain stable and shrimp exports are not reduced. Projections by a Fisheries Sector Review and Future Development (FSRFD) study suggest medium term growth rates of 5.6 and 3.2 percent per year for inland and coastal aquaculture,respectively (Table 2.5). 2.21. Inland aquaculture is the largest source offish production amongst the variousfisheries sub- sectors,production in 2002 stood at an estimated 850,000 tons with over 80percent from small holder- pond polycultures. (Other areas of importance include seed production, pen and cage culture and commercial aquaculture.) New ponds continue to grow at a rate of 2 percent or more per year. Small pond carp polycultureswill continueto offer the greatest potential for expansion and increasedproduction within the sector. Fresh water shnmps (golda) are cultivated by about 105 thousand farmers, with total production of about 10thousand tons;' about 1percent of total inland aquacultureproduction by weight. Integratedrice-aquaculture systemshave proven highly productive, although on an, as yet, limited scale. 2.22. Brackish-water aquaculture produces shrimp, prawn and fish, and is expected to double its small base (almost 95,000 tons in 2002) over the next decade, although much depends on the ability to meet SPS requirements of importers. Tiger shrimp (bagda)produced by this technique comprise most of Bangladesh's shrimpexports. 2.23. There is a strong need and, fortunately, significant potential remains for improving productivity in shrimp production. Since total pond area is not expected to increase significantly, major increases in total shrimp production will have to come through increased productivity per hectare. Currentproductivity in Bangladesh shrimpproduction (100-150 kgshectare) is far below that of Thailand (4-7 tonsihectare), the leading shrimp e ~ ~ o r t e r .Raising~ productivity is possible through increased ~ " ~ support of extension, improved pond construction, better hatchery and water management, reduced Post Larvae (PL) mortality, and improved stocking ratios. Most important are improvements in post harvest handling to raise quality and safety of the final products, issues of major concern to international buyers (see Annex Table 2.1). 30Collis (2004),p. 89. 31Dobson and Quader, 2004. 32Collis (2004) reports a somewhathigher average yield of 214kgshectare in 2001102 (See Table 9). According to the Bangladesh ShrimpFoundation, yields in Bangladeshare about 250 Kgha compared to 1000Kg/Ha in India. 2.24. Key issues in the sub-sector - particularly for poorer farmers - are (i) availability of information, (ii) extension support and (iii) quality of available fish seed and access of the poor to production units. Landless and marginal farmers and the fish seed industry will require additional investment and technical support. Government-NGO partnerships have proven to be the most effective means for promoting and improving smallholder aquaculture. Capture Fisheries 2.25. Open water capturefisheries -a major income sourcefor the rural landlesspoor - have been in decline in recentyears. Catches of the critically important hilsa, Indian Salmon, the major carps (Rui, Catla, Mrigal and Kalabaus) among others have declined precipitously. Production losses in inland capture fisheries have been on account of habitat loss, reduced dry season flows and unregulated over- fishing as wetlands and other aquatic resources are lost to infrastructure, agriculture, urbanization and increasing pollution. Portions of the marine capture fisheries -inshore coastal and tiger shrimp capture - are also experiencing a decline, which is especially worrisome as the sub-sector is a primary source of income and nutrition for at least half-a-million households in the coastal region and also employs about 0.4 million poor in seasonal shrimp larvae collection. 2.26. Key issuesfor the sustainable use andpreservation of the capturefisheries include maintenance of dry season flows, protection of key aquatic habitats, creation and maintenance of fish sanctuaries, limitation to further disruption of migratory pathways and avoidance of further pollution and chemical degradation. In addition, the future of the sub-sector will require a reduction of fishing effort. Recent small scale efforts have shown that communities working with local governments are capable, with technical and other support, of managing local resources in a sustainable manner. Livestock and Poultry 2.27. There has been virtually no growth in the number of large animals (mainly cattle and dairy cows) since the early 1980s. This is a significant loss to the rural economy. High cost of feed and fodder, and productivity and animal losses from disease are major constraints on production. In addition, shortage of veterinary personnel and equipment to prevent and treat infectious diseases contributes to high mortality rates of animals.33 Increasing the number of improved breed cattle appears to be a priority for raising productivity in the sector. 2.28. Theformal sector commercial dairy market is growing rapidly and increasing its relatively small market share, while the markets for meat remain almost entirely in the informal sector. As recently as 2000, the informal sector accounted for more than 95 percent of total milk marketed. In recent years, however, formal sector enterprises have established chilling centers, and milk collection networks linked to village milk co-operatives. BRAC Dairy now sells 40-45,000 liters of milk per day, equivalent to 20 to 25 percent of the commercialmilk market.34 2.29. In contrast to large livestock, thepoultry sector, particularly chickens and eggs, has seen good growth. Chicken meat and egg production each rose by 6 percent per year between 1986 and 1996. Figures for later years vary widely according to source. FA0 data indicate a slowdown in growth since 1997, while BBS figures indicate an average annual growth rate of 6.7 percent from 1995 to 2000. Estimates from the Bangladesh poultry association suggest growth rates closer to those of BBS. Urbanization and rising incomes are the major forces driving increased demand for poultry products in 33Control of disease is complicatedby the unregulated cross border import of livestock from India (Alam (1997)). 34Sales of other processed products such as yoghurt and flavored milk are also increasing rapidly, and account for about 20 percent of BRAC Dairy's sales. (Source: interviews with BRAC Dairy, 2005). Bangladesh. Since most commercial demand is concentrated in major urban centers, poultry production on the approximately 65,000 commercial farms is largely a peri-urban activity. 2.30. Indications are that expansion of the poultry industry may be slowing down, as it confronts lingering concerns. According to the Bangladesh Poultry Association, operators of about 800 - 1,000 medium-sized poultry farms have appealed to the government for assistance with servicing their outstanding debt from commercial sources. Demand fluctuations (related to consumer fears surrounding the Asian bird flu) are reported as major reasons for somerecent failure of poultry operations. 2.31. High costs of maize and soybean meal have also slowed growth in recent years and also threaten the competitiveness of the poultry sector in the medium-term.j5 Informal estimates by the Poultry Association suggest that imports (about 60 percent from India) account for about half of total annual use of maize as poultry feed.36Partly as a result of the transportationand marketing costs involved in importing, maize prices in Bangladeshare about 30 percent higher than in India and also higher than in other maize exporting countries,putting domestic poultry producers are at a disadvantage. It is estimated that higher costs of feed add about 20 percent to the production cost of layers and 8 percent to the production cost of broilers relative to the producers in India.37 HI. The Rural Non-Farm sector3' 2.32. The rural non-farm sector accounts for more than 40 percent of all rural employment and more than half of rural household income, and is growingfast. The RNF sector here is defined as all activities located in villages, rural towns, and peri-urban areas excludingthe primary production of crops, fish, trees, and livestock. The number of non-farm enterprises increased from about 2 million in FY90 (BBS Integrated Annual Survey of Non-Farm Activities (IASNF)) to about 4 million in 2003 (National Private Sector Survey of Enterprises in Bangladesh (NPSEB)). Annex Table 2.4 describes the main characteristicsof the micro, small, and medium enterprises,including those in the rural areas. These data also suggest that employment in rural non-farm activities increased from 5 million to 9 million over the same period. As in many other developing countries,wholesale and retail trade is the most common non- farm activity, with about 57 percent of rural firms engaged in this sector. Another quarter of the enterprisesin rural areasare manufacturing firms. 2.33. A number of key investment climate weaknesses are holding back the growth potential of the rural non-farm sector. The micro, small and medium (MSM) enterprises in non-farm sector, (which constitute almost all of the rural non-farm enterprises) appear to be lacking access even to basic factors of production and services (Table 2.6). Only 31 percent of rural enterprises reported having utility connections, compared with 60 percent in urban metropolitan areas.39Moreover, only about 2 percent of the MSM firms have own generators40and only 1.5percent have access to a phone. About one quarter of rural enterprises said they had access to formal credit, which was significantly higher than for the urban areas. The widespread and rapidly growing availabilityof micro-enterprise lending by NGOs appears to be the drivingforcebehind the higher access in rural areas. 35High cost of medicines and vaccines imported fiom the European Union and the Unites States also contribute to high costs of poultry production in Bangladesh. 36Note that officialdata seriously under-report maize imports. 37See Quasemand Islam (2004),p. 348. This section is based on Promoting the Rural Non-Farm Sector in Bangladesh, World Bank, 2004. 39The estimate of utility connection may be an under estimate as it is derived fiom responses of the enterprises about their expensesfor utilities. 40This comparesto 72%of in the Investment Climate Assessment (ICA) Survey, 2003, which collected information on larger firmslocated in Dhaka and Chittagong. Table 2.6: Bangladesh: Access to Services and Credit by Micro, Small and Medium Enterprises Rural Other urban Peri-urban Metropolitan Total Enterprises with Access to Services (%) Utility Connection 31.49 59.65 57.05 59.88 40.00 Generator 1.81 3.96 3.40 4.15 3.00 Phone 1.51 5.94 4.54 9.79 4.00 Registered 13.20 37.93 22.05 41.78 21.11 Association Member 16.21 25.08 16.04 21.95 18.41 Access to Credit (%) None 41.25 47.03 54.62 49.06 43.74 Informal 22.23 20.96 27.23 32.86 23.71 Formal* 24.62 20.02 9.89 11.02 21.45 Both 11.90 11.99 8.27 7.06 11.10 *Formal includes micro-lending by NGOs. Source: the Rural Non-Farm Sector in Bangladesh, WorldBank, 2004; WBStaflestimates based on NPSEB (2003). 2.34. Average labor productivity of rural MSM firms is estimated at only about 60percent that of urbanfirms. Estimates of total factor productivity (TFP) are broadly similar across regions and sectors, but difference among firms within a particular region andforsector spear to be large. Analysis shows that access to markets, access to services and toformalfinance, education of theproprietor and access to a networkfor contract enforcement are all statisticallysignificant determinants of TFP, after controlling for other firm characteristic^.^' It is womsome, therefore, that more than half of the firms covered by the NPSEB reported lack of access to finance as the biggest problem in starting an enterprise, and about 44 percent of firms reported it as the biggest problem in running an existing firm. Floods and disasters, availability of electricity, and road conditionswere also identified in the NPSEB as severeproblems by at least a third of respondents. IV. Implicationsfor a Rural Growth Strategy 2.35. Agricultural growth remains crucial for rural poverty reduction in Bangladesh, and rice production will continue toplay an important role. The development strategy should neither neglect nor over-emphasize the rice sector, however. Bangladesh has a comparative advantage in rice production, particularly in the aman season, when there are few profitable economically viable alternatives to rice cultivation on flooded land, the economic potential of which needs to be exploited. Even though urbanization and related shifts in diets are mitigating factors, growth in overall population and rural incomes can be expected to lead to increased demand for rice. Background projections for this report suggest that total demand for rice is likelyto increaseby 30 to 40 percent over the next two decades, even with constant real prices. The emphasis would have to be on improving productivity, which would also free up resources for use for non-rice For this, expanding the use of the H W variety of rice to the aman crop (where H W is being used in only half of the area) holds most promise. 41For fiuther details on the TFP calculations see Promoting the Rural Non-Farm Sector in Bangladesh, World Bank, 2004, Vol II., pp. 46-48. 42Development of shorter-season rice varieties and greater use of mechanization could permit expansion of non-rice crop area and production (mainly in the winter season) while maintaining adequate domestic rice production. Shortage of water in the winter season and problems of flooding and drainage in the monsoon season are perennial issues that remain unresolved in many regions of the country. Increasingly frequent localized water shortages (declining water tables) in the winter season add to the costs of irrigation and reduce the profitability of boro rice in some regions, favoring the production of less water-intensive crops. 2.36. Ultimately, raising real agricultural incomes in Bangladesh will require more diversifiation into higher-valued crops and increased output of the fishing, poultry, and livestock sub-sectors. Analysis of comparative advantage suggests that in the winter season, production of most non-rice crops (including wheat, pulses, potatoes and many vegetables) is economically efficient.43 Marketing constraints including shortage of cold storage facilities and a functioning cold chain, lack of grades and standards, insufficient processing capacity and, to some extent, weak consumer demand related to low incomes and consumerpreferences,remain a major constraint, however. Transport bottlenecks also slow delivery of fresh vegetable products and lead to reduced quality and sales price. Promoting development of more vertically integrated marketing systems through NGOs (e.g. BRAC), private firms (including supermarkets) through improvements in road and electricity infrastructure, and in local governance can help spur the necessary private investments. Medium-term prospects for the poultry sector and the commercial dairy sector are also good, benefiting from rapid urbanization and rising per capita incomes. The country needs to be vigilant against outbreaks of poultry diseases (including the Asian bird flu), and provide improvedveterinary services,in general. 2.37. Bangladesh appears to have a long-term competitive advantage in inland aquaculture. Even if past declining trends for inland and marine capture fisheries continue, as is expected, fish production as a whole can still be expected to grow as long as inland and coastal aquacultures continue to do well. In addition to research, information, and extension support from the Government, growth in aquaculture will need improvements in post-harvest handling and marketing and distribution of better quality fish seeds. Coastal shrimp and prawn aquacultureare typically profitable and practitionersare, for the most part, able to finance their own activities-special attention is needed, however,to protect the rights of small farmers and landowners, particularly in coastal areas. Although much of the direct income from production accrues to the land-owners, there are still major pro-poor employment and income effects in the provision of inputs (e.g. capture of wild shrimp post-larvae (PLs) by poor households) and processing. Shrimp exporters also have a role to play.44 First, they will need to adhere to self-imposed standards and the recognition and acceptance of third-party certification process by importers and regulatory agencies in importing countries. Second, in the medium-term, farmers need to adopt semi-intensive and intensive methods of shrimp farming in place of the prevalent practice of extensivepractice. 2.38. The Government's role in promoting agricultural diversification is important, albeit limited. In particular, a significant expansion in agricultural research and extension efforts is needed in rice as well as high value added crops, Iivestock andfish. Bangladesh currently spends only about 0.2 percent of agricultural GDP on agricultural research, compared to an average of 0.6 percent of agricultural GDP for all developing countries. Given the importance of agriculture and also the potential for future productivity gains, a medium term target of 2.0 percent of agricultural GDP seems appropriate. Establishment of an endowment Trust Fund, and institutionalreforms to increase stability of funding and to strengthen research management, (including renewed efforts at prioritizing research) could substantiallyincreasethe effectivenessof public research and extension e~~enditures.~' 2.39. In addition to agricultural research and extension, public investments in rural marketing infrastructure (particularly roads, bridges and telecommunications) combined with a favorable environment for private investment (e.g. in warehouses and cold storagefacilities) are needed. This would reduce storage losses and price risk for non-rice crops, particularly highly perishable fruits and vegetables, and improve farmer incentives for expanded production. Government also has an important 43Shahabuddinand Dorosh (2004). 44Bangladesh: Growth and Export Competitiveness, World Bank, 2005. 45Revitalizing theAgricultural TechnologySystem in Bangladesh. Bangladesh Development Series (paper no. 7), World Bank. 2005. Dhaka, Bangladesh: World Bank office. role in providing a growth-friendly institutional environment "rules of the game" for ago-food system and agro-enterprisedevelopment (see Box 2.1 for relevant lessons from a recent study for India). Box 2.1: Role of Government in Agro-Food System and Agro-enterprise Development Setting and ensuring enforcement of transparent and consistent 'Rules of the Game' Establish and enforce rules which define and allocate property rights (i.e. property and bankruptcy laws; intellectualproperty rights; zoning regulations). Establish and enforce rules which define permissible and non-permissible forms of cooperation and competition(i.e. licensing laws, laws of contract and liability, company and cooperative laws; anti-trust laws). Establishand ensure compliance with bio-safety, food safety, worker safety, and sanitationregulations. Negotiate favorable terms for access to international markets and ensure fair practices on the part of internationaltrading partners. Addressing market failures Ensure that the country is protected fiom the harmful introduction/spread of plant pests and animal diseases. Ensure the availability of (production, price, industry) information and statistics to facilitate market activity and to monitor market progress. Invest in or facilitate risk management instruments for agribusiness system participants (e.g. futures contracts,options, negotiable warehousereceipts, crop insurance). Compensate for unbalanced power relationships within the agribusiness system by monitoringpotential abuses of market power, by providing training and information, andlor by supporting organizational developmentamong weak participants. Compensate losers in structural reform processes through safety nets and other transitional targeted programs. Building physical and knowledge capital Invest in social overhead infrastructure,especiallythat related to transport, and energy. Invest in knowledge-building to accelerate the agribusiness learning process and better enable the emergentprivate sector to participate/compete(i.e. R&D; academic/technicaltraining). Facilitatedevelopment of agriculturalmarketing facilities(i.e. marketplaces;wholesale markets). Source: India Re-energizing the Apicultural Sector. World Bank. 2004. Report No. 27889-IN. 2.40. Improved marketing infrastructure will also promote growth of non-crop agriculture, including thepoultry andfish, as well as the rural non-farm sector. Meeting the growing demand for meat and eggs deriving will require steady increases in the supply of day-old chicks and feed (currently largely based on imported maize). Extension efforts are needed to ensure that poor rural households have access to technology in production and processing.46 2.41. Policies and investments outside the agricultural sector are also key to promoting non-farm growth. There is an urgent need to improve rural access to finance and electricity. Although, large farmers have access to formal credit and poor households can often obtain micro-credit, small and medium farmers have difficulty obtaining credit for production activities. Similarly, small rural non-farm enterprises report access to credit as a major constraint to establishing and operating a firm. Furthermore, the experience of Latin America and the Caribbean suggests that improvements in education, transport infrastructure, and engines of non-farm growth that promote female employment can be important components of a successful strategy to promote non-farm growth and rural poverty red~ction.~' 46Sen (2003). 47Berdegue, Reardon, Escobar (2000), and India Re-energizing the Agricultural Sector, World Bank. 2004. Report NO.27889-IN. Chapter3: The Labor Market: Trends, Characteristics, and Challenges I. Introduction 3.1. Labor it is ofien said is Bangladesh's mostprecious resource. This precious resource, under the right conditions, has the potential to unleash a rapid transition toward prosperity, as the East Asian economieshave shown without having the benefit of a natural resource bounty. A relative abundance of labor provides Bangladesh with a comparative advantage in production of labor-intensive goods. Of course, global comparative advantage does not automaticallytranslate into global competitiveness,which depends on a host of factors that affect productivity. Productivity in Bangladesh does suffer fiom a number of problems that are highlighted in the other chapters of this report. Are there inefficiencies in the labor market itself which undercut overall productivityof the economy? This chapter assessesthe key labor market trends and characteristicsin seekingto answer this. 3.2. Creating productive employment opportunities for a rapidly growing labor force clearly is a major challengefor the Bangladeshi economy. Its substantial labor force of about 50 million is growing at a rapid 3 percent per year - adding just under 2 million workers each year.48 The rapid labor force growth derives from two main sources: the demographic transition which is resulting in a surge of youthful entrants into the labor market, and; the sharp increase in female labor force participation since the mid-1990s. Expectedly, given Bangladesh's low development level, almost half the labor force is still in the agriculture sector and three-quarters are in rural areas. At the same time, the labor force appears to be in the midst of transitioning fromrural to urban areas, consistent with the overall economic shift that is discussed in Chapter 5, and from agricultureto rural non-farm activities. 3.3. Despite the considerablepressures, labor market outcomes have been generallyfavorable. For one, most new entrants between 2000 and 2003 (the years in which the latest available Labor Force Surveys were conducted) were able to find a job, keeping the unemployment rate at about 4 percent. However, this is slightly tempered by the fact that the entrants were partially accommodated by sharing some of the existingjobs, as indicated by a decline in the average hours worked per week per worker in all economic sectors during 2000-03. Rising real wages across the major economic sectors provide further evidence of labor market tighteningdespite the rapidly growing labor supply. 3.4. At the same time though, some important aspects of the labor market deserve closer attention. The unemployment rate is 8 percent for the 20-24 year olds and even higher for those with higher secondary or degree education. The income levels of the self-employedworkers (who account for almost half the workforce) appear to have declined over the 2000-03 period. Furthermore, the formal sector comprises only about a fifth of the labor force (although this ratio is higher than in India) and shows no signs of piclung up - in fact, data for the 2000-2003 period show a decline in the formal sector's share. The share of "unpaid family workers" in the workforce increased over the 2000-03 period, at the expense of all otherjob statuses except "employers". 3.5. A particular concern afflicting the labor market is acute skills shortages. Not only are the education levels among the workforce low, the education and vocational training system are not geared toward the market needs. On average, the employed labor force has only 4 years of schooling. It is a major loss to productivity when even this scarce human capital is not properly utilized, as highlighted by the fact that almost one-fourth of the unemployed hold a SSC, HSC or higher degree. Furthermore,just over 1percent of the labor force has had technical training or vocational education,and even among those 48The 2002-03 Labor Force Survey (LFS) estimated the labor force size at 46 million. It also estimated the annual increase in labor force at over 1.5 million, which would imply a labor force of over 50 million today. the relevance of their skills is questionable. The low quality of the vocational education and training (VET) system is reflected by its low capacity utilization and pass rates and also by the weak demand for its graduates. A tracer study shows that less than 10percent of VET graduates were employed over two years after the survey, while close to half the graduates took at least a year to find a job, and this in an economy where slulls are deemed to be acute shortage. Not only that, wages of VET graduates apparentlyare lower than those of graduates fromthe general education system.49 3.6. Other key identifiabledistortions in the labor market include: There is a clear segmentation of the public and private job markets - this is apparent in the considerable wage premium (over 30 percent) for public jobs, after taking into account worker characteristics such as education, skill, gender, etc. Analyses have shown the disparity to be sharper at the lower-end (classes 111 and N)public sector jobs. Not only does this lead to rationing of the high premium public sectorjobs, it also distorts wage signalsacross the economy. Another segmentation is between urban and rural areas: Econometric analysis in this chapter shows that urban areasprovide more employmentopportunities (more hours of employment)and a significant wage premium. The baniers to rural-urban mobility of labor - that would also be productivity improving - likely have to do with the shanty living conditions in the cities, especially for fresh migrants and the poor, and the higher cost of living. An urban agenda of the sort discussed in Chapter 5would work to improve this. Female workers continue toface considerable barriers. Improvementsnotwithstanding, only a quarter of working-age females participate in the labor force and wages for female workers are significantlylower than for male workers with similar characteristics. This emanates mostly from the social baniers commonly seen in developing countries, but it is hard to imagine a path of rapid development, if three-quarters of females are not there to participate in and shape it. The solutionsof more educationand female empowerment are happening but could be speeded up. 3.7. On balance, labor laws and regulations do not appears to be serious hindrances to the functioning of the labor market, although worker rights need better protection. Laws and rules are relatively light and, if anything, suffer from ineffective enforcement - certainly there are no signs of the draconian labor regulations seen in India. Overall, firms retain considerable flexibility to hire and fire, as confirmed by the Investment Climate Assessment and Doing Business survey results. At the same time though, lack of organized union activity in the formal private sector, where only about one worker in twenty is covered, also has its down sides, especially since it arises from government restrictionswhich the ILO's Committee of Experts on the Application of Convents and Recommendations has deemed excessively restrictive. Such limited union activity handicaps workers' ability to engage in collective bargaining and protect their rights, and hinders the formation of an effective mechanism to resolve labor disputes. The recent impasse in the RMG sector over worker demands for higher minimum wages highlights the costs of the latter problem. It is not clear though whether the resolution to this is government interventionand, if at all, in what shape and form. 11. Labor Market Trends in Bangladesh Long-term Demographic Trends: Transition to a Younger, More Gender-Balanced, and More Urbanized Population 3.8. Bangladesh's population has almost tripled since 1960 - increasing from 52 million inhabitants to over 140 million. Female population has grown fasterthan male population, reflecting the impressive underlying improvements in gender parity, with the share of female population in total 49The Bangladesh VocationalEducation and TrainingSystem: An Assessment, World Bank, 2006. 29 population increasing from 47 percent in 1960 to 49 percent in 2005, slightly higher than in India and Palustan. Bangladesh has experience the biggest increase in the ratio in the South Asia region since 1960 (Table 3.1). This is an impressive, if somewhat underreported, achievement that benefits from increasing empowerment of females in the social sphere - female primary enrollment rates are higher than male enrollment rates, for example. The rapid urbanization of the population is discussed in Chapter 5. Source: WorldDevelopment Indicators, WorldBank 3.9. The growth of the working-age population has outpaced that of the population as a whole. While total population has increased by about 70 percent between 1980-2005, worlung-age population (those in the 15-64 age group) practically doubled - increasing from 44 million in 1980 to 86 million in 2005. Demographic factors such as increased life expectancy and markedly lower infant mortality have offset the slowdown in population growth (from 2.5 percent in the 1980s to about 1.5 percent), resulting in working-age population growth remaining largely unchanged at about 3 percent per ann~rn.~'Under-5 mortality rates fell from 205 per 1,000 live births in 1980to 77 in 2004; one of the fastest rates of decline in the developing world." Life expectancy at birth increased from 49 years to 64 years over the same period. The labor dividends from the ongoing demographic transition are expected to continue for a considerable, although at a declining rate. The growth rate of worlung-age population is expected to slow down to 2 percent a year in the coming years. Still, more than one-third of the population is currently under the working age, a sizable pool to feed robust labor supply growth. Under certain baseline assumptions, the working-age population is expected to increase by 65 percent by the year 2040.'~ Figure3.1: Age-Wise Demographic Trends in Bangladesh t Populationaged 15-64 xPopulationaged65+ I Source: WorldDevelopment Indicators, WorldBank 50Working-age population estimates vary by source. While the World Bank's WDI data show average 2.7% growth during the 1990s(virtually the same as in the 1980s), the ILO and the US Census Bureau each report the growth rate at about 3.4 percent, about one percentage point higher than the average growth reported by the ILO for the 1980s. 5' Even so, Bangladesh still has long ways to go. Just to meet its MDG, Bangladesh needs to lower the U-5 mortality rate to 48 by 2015. After that a target rate of 15-the current rate of Sri Lanka - or less can be set. 52The precise demographic assumptions - regarding birth and death rates, population growth, fertility rates, infant and child mortality, and life expectancy - can be found in the World Bank's SIMA database. Labor Force and Employment Trends A rapidly r~rowin~labor force 3.10. The 2003 LFS estimated the size of the econonzically activepopulation, or the laborforce, to be 46.3 million - or about 57 percent of the working-age population (Figure 3.2.a).53 Reflecting the predominantly rural population in Bangladesh, 76 percent of the labor force resided in rural areas. The female labor force was estimated at about 10 million, or 22 percent of the total labor force. Consistent with the demographic trends, survey results showed the labor force to be quite young, with over half the workers falling in the 15-34age group (Figure3.2.b). Figure 3.2.a. Labor Force Trends Figure 3.2.b. Age Profile of Labor Force Source: Bangladesh Labor Force Surveys (variousyears) 3.11. The labor force has grown at a fast pace, reflecting both the rapidly increasing working-age population and, more recently, sharp increases in thefemale laborforce participation rate (LFPR). As the working-age population increased by over 11 million between 1996-2003,'~the labor force increased by over 10million (Figure3.2.a); 4.6 million between 1996-2000and another 5.6 million between 2000- 2003, with average annual labor force growth of 3.6 percent over the 1996-2003period. Consistent with the overall population patterns, the labor force is also becoming increasingly urbanized - urban labor force's share in total increased from 23 percent to 24.5 percent over 1996-2003. 3.12. Thefemale laborforce almost doubled in size between 1996-2003 to 10 million, while the male labor force grew by just 17percent. The female labor force grew at an astounding annual rate of 14 percent over 1996-2000, which then slowed down a bit but still remained high at 7 percent during the 2000-2003 period. The LFPR increased sharply for females in urban and rural areas, and in both areas the increase was more pronounced in the 1996-2000 period than the 2000-2003 period (Figure 3.3.a). Almost 4 million additional female workers entered the labor force during 1996-2003 due to the increase in the female LFPR alone. Nonetheless, because of the low starting base, the female labor force remains less than a third of the male labor force of 36 million. In the meanwhile, the male LFPR remained virtually unchanged over 1996-2003. 53The LFS also counts those aged 65+ years as being part of the labor force. This group, with 2.3 million persons, comprisesjust 5% of the total labor force though. 54The increase in the population aged 15+ during the 1996-2003 period was 11 million as per the LFS and 15 million as per the World Development Indicators. Figure 33.a Changes in LFPR Over Time Figure 3.3.b: Age-LFPR Relationship Source: Bangladesh Labor Force Surveys (various years) 3.13. The age patterns of LFPRs differ considerably between male and female workers (Figure 3.3.b). The male age-LFPR relationship has an inverse-U shaped pattern. The youngest and oldest male workers have the lowest LFPR. LFPR for males in the 30-54 age group was close to 100 percent and over 97 percent for the 55-59 age group. Two-third of males work even after the age of 64. LFPR for those in the 15-19 year age-group was only 50 percent, reflecting in part school enrolment. The LFPR pattern is almost identical for rural and urban males, with the LFPR of urban youth being lower, reflecting differences in school enrolment. The age-LFPR relationship for female workers shows a differentpattern and also some urban-rural variation. For one, the female LFPR is significantlylower than the male LFPR for all age groups. The urban female LFPRs are higher for all age groups except the 15-19 age group, as girls in urban areas are more likely to be enrolled in school. LFPRs for both rural and urban females drop precipitously from the 15-19 age group to 20-24 age group, from about a third for urban females and 40 percent for rural females to around one-quarter for both, most likely because of marriage. For rural females, the LFPR then remains virtually constant till age 39, while for urban females it increases a bit in that interval. The LFPRs for both groups decline from the 40-44 age group onward. Employmentand underemployment trends 3.14. The Bangladeshi economy added 5.3 million (9.5 million) newjobs over the 2000-2003 (1996- 2003) period. This is a sizable number, although it still falls short of the 5.6 million (10.2 million) new entrants into the labor force. As seen in Figure 3.4, new jobs for females kept pace with the increase in female entrants in the labor force during the 2000-2003 period, whereas half-a-million new male entrants into the labor force were not accommodated. A further breakdown reveals that 2 million new jobs were created in urban areas, where the labor force increased by 2.1 million, while 3.3 million new jobs were created in the rural areas, where the labor force increasedby 3.5 million. Figure 3.4: Job Creation Vs. New Labor Force Entrants Source: Bangladesh Labor Force Surveys and WorldBank Staflcalcz~lations 3.15. Unemployment rates in Bangladeslz, as in the majority of the LICs, have been consistently low. Standard definitions of unemployment consider any person of age 15 years or above as unemployed if he or she did not work at all during the preceding week of the survey and was actively loolung for work. As per this definition, 2 million persons, or 4.3 percent of the labor force, were unemployed in 2003 (Table 3.2). The unemploymentrate increases with the education levels of the labor force,reflecting the fact that the more educated workers have higher reservation wages and are less willing to compromise on the quality of the job they accept. Also, those who go for higher studies often come from more well-off backgrounds and can afford to hold-off working longer. At 8.3 percent, the unemployment rate was the highest amongthe 20-24 age group. Table 3.2: Unemployment rate among labor force by level of education and sex Bangladesh Urban Rural Both Male Female Both Male Female Both Male Female sexes sexes sexes Per cent equivalent 1 Degree & 9.5 8.1 17.4 7.7 6.2 14.7 12.0 10.6 22.8 above Others 9.6 8.9 15.7 9.0 5.7 31.2 f10.0 11.0 0.0 Source: Bangladesh Labor Force Surveys (variousyears) 3.16. The open unenrployment estimates based on tlre standard definition of unemployment besides being exceptionally low do not change nruch over time and as such are not very informative. By focusing on only those who did not work at all, the unemployment indicator misses the inefficiencies in the labor market that affect the majority of the labor force who are looking for more substantial work hours. For this reason, the underemployment rate is considered a more meaningful indicator of the labor market condition. 3.17. Unfortu~zately,data quality and dejinitionalproblenzs preclude a meaningfkl assessment of the underemployment trends in Bangladesh. The LFS considers as underemployed anyone who is part of the labor force but did not work for at least 35 hours a week, regardless of whether or not theperson was willing to workfor 35 hours or more. Furthermore, comparability between the 2000 and the 2003 surveys is rendered difficult by the fact that while the former was conducted on a rotating basis covering the entire year, the latter was conducted only during the months of October and November - a slack period for agriculture, which would result in overestimation of underemployment. Therefore, even though the 2003 LFS reports an increase in underemployment over the 2000-03 period, from 16.6percent in 2000 to 37.6 percent in 2003, it is not at all clear whether this is reflects actual increase rather the seasonality in data collection andlor the fact that female workers - who are increasing their share in the labor force - ofien tend not to look for full-time work to balance competing work pressures at home. 111. Employed Labor Force Characteristics Sector composition of the employed labor force 3.18. The agriculture sector has the highest share of the employed laborforce, as in other most other low-income economies. In 2003, despite accounting for just 21 percent of GDP, agnculture sector accounted for 52 percent of the labor force, up from 49 percent in 1996 (Figure 3.5). The increase in agriculture's share over 1996-2003,as suggested by the LFS results, is counter-intuitive though as it came at a time when agriculture's share in GDP fell from 25 percent to 21 percent. It is also in sharp contrast with recent analytical work (based on a separate survey of rural population) that shows a rising role of non-farm activities in rural employment.55 On balance, the evidence is more compelling that the employment share of agriculture has been declining over time, as would be expected from the sector's diminishingrole in the overall economy. Figure 3.5: Distribution of employmentby broad sector: 2003 10% Source: Bangladesh Labor Force Surveys and World bank Staff calculations 3.19. Gender decomposition of labor across sectors provides additional insight into the sectoral mobility of labor. Almost 90 percent of the male workers are either in agriculture (50 percent) or services (38 percent): the share of male workers in industry (manufacturing and mining) and construction, which combine for a quarter of the GDP, is thus only about one-tenth. Their share in agnculture declined over the 1996-2003 period (Figure 3.6), while the share of the female workers increased substantially - from 28 percent in 1996 to 59 percent in 2003. Again, these trends derived from the LFSs are likely overstating the increasing trends of female participation in agriculture. Nonetheless, these trends may well be highlighting the lower mobility of female labor out of agriculture than of male labor. That would 55 Mahabub Hossain (2004). suggest that the positive aspect of increasing female labor force participation is, at least partially, being offset by the fact that their entrance is mostly in the agriculture sector (including livestock production), which is not only the least productive and remunerative sector in the economy but in which women's involvement usually takes the form of unpaid family work which does not give them direct access to income and may not be a particularly effective means of empowerment. Figure 3.6: Gender Decomposition of Labor Across Sectors Source: Bangladesh Labor Force Suweys (variousyears) 3.20. A more detailed breakdown of the labor force by industry (4 digit BSIC level) is shown in Annex Table 3.3. After agriculture, wholesale and retail trade is the next most significant category, employing about 14 percent of the labor force, while manufacturing accounts for one-tenth. There are some interesting gender and urban-rural difference within these two categories. Wholesale and retail account for 21 percent of the workers in urban areas, which is twice the category's share in rural areas. It also accounts for 26 percent of urban male workers and 14percent of rural male workers, which, in each case, is about eight times the category's share of female employment. It is quite telling that, despite the significant decline, manufacturing sector's share of female employment - in both urban and rural areas - is more than twice its share of male workers. For the urban sector at least, this in large part reflects opportunities for female workers in garments manufacturing. Female participation in rural manufacturing deserves more analysis. Status of employment 3.21. The seljlemployed workers constitute the largest group, by far, of any employment status category (Table 3.3), accountingfor 45percent of the employed laborforce. 56 This is true not only for rural areas where family based subsistence agriculture has been traditionally the main economic activity but also for urban areas where it is likely being driven by the large share of male employment in wholesale and retail trade. Another 18 percent classify as ztnpaidfamily workers: thus family or own source of employment contributes about two-third of the labor force. The breakdown by gender for these two categories is informative. Almost half the male workers classify as self-employedwhereas only a quarter of female workers do. On the other hand, almost half the female workers classify as unpaid family workers, which is up from 19 percent in 1996 (Annex Table 3.5), while only one-tenth of male workers do. Among the 34 percent who classify as hired workers, employees are more common in urban 56 Self employment as a percentage of non-agricultural employment (data for the latest available year between 1990-2000) in India and Pakistan was 37 and 43 percent (HDC 2004), respectively, compared to 44 percent in Bangladesh. Share of paid workers (regular and casual) in total employment in India were 37 and 46 percent respectively among men and women, compared to 35 percent and 20.0 percent in Bangladesh. 1995-96 data (HDC 2004) shows that agriculture contributed 67, 47 and 42 percent of employment in India, Pakistan and Sri Lanka respectively. The comparable figure for Bangladesh was 48 percent. areas and day laborers in rural area. As would be expected, regular salaried employeejobs are mostly in the non-agriculture sector while the agriculture sector mostly creates jobs in the categories of self- employed, day laborer, and unpaid family workers. Table3.3: Employed person by status of employment, gender and locality (Percent) I I Bangladesh I Urban I Rural I Status of employment Source:Bangladesh Labor Force Survey 2003 3.22. The income and poverty situation of self-employed workers in rural areas may have worsened during 2000-2003. If we consider monthly income of taka 2,000 (roughly 1 US$ a day) cut-off for poverty, about half of the self-employed fell below this level in 2003 (Annex Table 3.10), up from 42 percent in 2000. About 35 per cent of the urban self-employed and 55 per cent of the rural self-employed fell below the cut-off point in 2003, compared with 40 percent and 43 percent in 2000. The worsening condition of the rural self-employed would be even more pronounced if inflation during this period is taken into account. It is hard to be conclusive about this worsening pattern, however, as the survey data on income from two LFSs are not strictly comparable. 3.23. The number of regular salaried jobs declined over the 2000-2003 period (Table 3.4), and the only reason this did not lead to increasing casualization of labor was that the day laborers declined even faster. The number of regular employees fell in all major sectors except industry where it saw a small increase. The number of day laborers on the other hand fell markedly in agriculture but increased in all other sectors. The reduction in salariedjobs affected the female workers much more severely -the number of female employees fell by 20 percent and the number of female day workers by 40 percent (Annex Tables 3.8 and 3.9). This was offset by an increase of about 75 percent for unpaid family workers and about 15 percent for self-employed. The number offemale unpaidfamily workers, infact, increased more than five-fold between 1996-2003, with most of the increase occurring in rural areas. The increase in unpaid workers was mostly accounted for by the agriculture sector and, to a much lesser extent, by the industrial sector. The unpaid family worker category,just 12 percent of total employment in 2000, accounted for two-third of the increase in employment between 2000-2003 and all of the increase in female employment. Table 3.4: Number of Workers by Status and Sector 2003 Data (thousands) Total Employee Employer Self-Employed Day Laborer Unpaid Worker Total 44322 6083 169 19849 8851 8139 Agriculture,Forest and Fisheries 22931 317 45 10548 5343 6555 Industry 4523 1666 24 928 1073 612 Construction 1541 87 6 341 1051 9 Services 15327 4013 94 8032 1384 963 Changebetween 2000 and 2003 (thousands) 11 kotal 5343 -450 72 1679 -639 3450 Agriculture,Forest and Fisheries 3147 -418 28 1994 -1888 3308 Industry 493 75 9 -280 259 210 Construction 446 -30 -7 35 426 -25 Services 1257 -77 42 -70 564 -43 11 The aggregate of changes in individual categories is less than the total change. This is because of addition categoriesin the 2003 LFS that were not there in the 2000 LFS Source: Bangladesh Labor Force Surveys (1999-2000 and 2002-03) EducationalAttainment of Labor Force 3.24. Improving the quality of the labor force, particularly its literacy levels, i~ crucial for an economy that wants to place itself on the path to development and modernity. Basic literacy and primary/secondary schoolingare essential to productivity of almost any sector, while good quality higher education plays an importantrole in the economy's transformationaway from primary activities. 3.25. In Bangladesh, the educational attainment trends of the labor force are somewhat disheartening. Half of the employed labor force is functionally illiterate and another 17 percent does not have education beyond class five (Table 3.5). On average, the employed labor force has only 4 years of schooling; 4.2 years on average for male workers and 3.2 years for female workers (Annex Table 3.11). The percentage of the labor force with no education, in fact, has increased over the period covered by the last three LFSs; from47 percent in 1996 to 50 percent in 2003. This is driven by the decline among male workers, urban and rural; female literacy ratios, both urban and rural, albeit still considerably lower than male has been stagnant at about 41 percent. Urban literacy rates (64 percent) are considerablyhigher than rural (45 percent), for both male and female workers. Table 3.5: Educational attainmentof the economically active labor force: 1996to 2003 Source: Bangladesh Labor Force Surveys (variousyears) 3.26. Education attainment varies considerably across sectors. The average education of employed persons disaggregated by sector, status of employment, gender and location are presented in Annex Tables 3.11-13. Agriculture (especially fishing) and transport sectors employ workers with the lowest average education, while the labor force in public administration, health and education services, and modem sectors like electricity-power, finance and real estate have substantially higher educational attainment. It is discouraging though to see that almost half of the workers in the manufacturing sector have no education, which likely adversely impacts the sector's productivity. 3.27. Datafor technical training and vocational education are similarly discouraging. Out of a labor force of 44 million, the 2003 LFS estimated that only 58,000 (53,000 males and 5,000 females), or a little over 1 percent of the labor force, had technical training or vocational education, and even among those, the relevance of their skills seems questionable. 3.28. There is still hope aspast trends in educational attainment of workers are expected to improve. With rising school enrolment, labor force entrants with school education tends to decline in the initial phase. This trend is likely to be reversed after the interval taken to complete school education. In fact, in Bangladesh, during the last three years there has been a small increase in the percentage of labor force with class IX and above education. It would be reasonable to expect that within the next few years the percentage of school educated labor force will increase because of the rapid increase in school enrolment since the early nineties. 3.29. A more intractableproblem has to do with the poor quality of education andpre-employment training. An important shortcoming is the gap between the skill types demanded by the market and those being supplied by the system is perceived to be still fairly wide. Studies have found education quality to have also deteriorated (CAMPE and UPL 2001). Furthermore, the quality divide between urban and rural areas is wide and, wonymgly, appears to be worsening. Addressing the urban-rural gap would have to be at the heart of any government plan to address the skills shortages in the economy. 3.30. The low quality of the vocational education and training (YET) system is reflected in the low capacity utilization andpass rates: in both vocational education and vocational training institutions, close to half the student capacity remains unutilized, and pass rates in standardized tests have also dropped significantly in recent years. This is further corroborated by the weak demand for the graduates from the VET institutions. A tracer study shows that less than 10 percent of individuals who graduated in 2003 from these were employed over two years after the survey, while close to half the graduates took at least a year to find a job. Moreover, the wages for VET graduates appear to be lower than those received by graduates from the general education system.57 Formal and informal employment 3.31. The informal sector is estimated to accountfor close to 80percent of the employed laborforce. Furthermore, the sector's share increased slightly over the 2000-03 period, accompanied by a decline in both private and public sector formal employment (Table 3.6). All economic sectors experienced an increase in informaljobs. Private formal sector job increased considerably in industries (manufacturing, mining, and utilities). These increases came at the expense of public sector jobs which declined across the board, and in private formaljobs in construction and services. 57TheBangladesh VocationalEducation and TrainingSystem: An Assessment,World Bank, 2006. 38 Table3.6: Formal and InformalEmployed Labor Force by Sector 2003 Data (thousands) IChange between 2000 and 2003 (thousands) Informal PrivateFormal Govt. [ Informal Private Formal Govt. Total Sector Sector Sector Total Sector Sector Sector Total 44322 35078 7368 1877 5343 5749 14 -419 Agriculture,Forest &Fisheries Industry Construction Services I I I Source: Bangladesh Labor Force Surveys (1999-2000 and 2002-03) 3.32. TheformaMnforma1 breakdown of employment needs cautious interpretation. For one, data on formallinformal distribution are difficultto interpret because of underlying definitionalproblems. Formal jobs are considered to be better in terms of job security, fringe benefits and pension schemes etc. Nonetheless, for an LIC like Bangladesh, it may be difficult to clearly distinguish formal from informal jobs and pinpoint their implications for worker benefits. In such a situation, even the formal sector employees may not get the benefits that are typically associated with the formal sector. Many formal sector enterprisesmay give jobs that are more closely associated with the informal sector. Besides, many employeesin the RMG sector in Bangladesh do not get an appointment letter even though they may work for enterprisesthat are registered and fall under the formal sector. On the other hand, many enterprises may not obtain the required registration etc. for qualifying as formal sector, because they do not have an incentive to do so. The distinction for the labor market in Bangladesh is made even more blurry by the fact that the LFS reports do not include a definition of the formal employment. Moreover, the workers may not know whether their employerhas formal registration when inteniewed for the survey. Bangladeshi migrant workers 3.33. International migration is a major source of gross national income andforeign exchange in Bangladesh. It is estimated that between 1976and 2004, a total of 3.7 million workers migrated abroad, and at present about 3 million people of Bangladeshi origin are living and working abroad (MOF).~'This constitutes6.5 per cent of the present labor force within the country (Annex Tables 15-16). The growth rate of fresh migration has declined since the early 1990s, with negative growth recorded for 2001 to 2003. Even so, 241 thousand workers -about one-half percent of the domestic labor force -migrated to work abroad in 2003. 3.34. The net benefis from migration appear to be positive for Bangladesh. The benefits from migration are clear - in the form of higher income and foreign exchange. Are there any associated costs, especially due to brain drain and shortages of critical skills? It seems unlikely that that is the case in Bangladesh, given its excess supply of labor, especially unslulled labor, a category under which majority of the emigrants fall. The share of the unslulled workers, in fact, has also risen since 1990 (Annex Table 17), from40 percent of all migrants in 1990to 54 percent in 2003. 3.35. The Middle East is thefavorite destination of emigrants - accounting for over 90 percent of emigration (Annex Table 18). Saudi Arabia alone accounts for almosttwo-third of emigration,but only about 40 percent of total remittances because of the low-skill emigrants it attracts. The U.A.E. and Kuwait are the other preferred destinations: together, these three countries account for almost 90 percent of all emigrants. The USA has a small share of immigrants from Bangladesh (less than 5 percent), but accountsfor 15percent of the remittancesbecause of the higher skill level of emigrantsthere. 58This does not take into account the estimatedmillions of unrecorded migrants, majority believed to be India. 39 IV. Earnings from Paid Employment Wage trends 3.36. Real wages in the manufacturingsector have risen quite sharply (Figure 3.7) as per the BBS data.59The real wage index for the sector increased by over 30 percent between 2000 and 2005. Real wages increased in the other sectors as well, albeit at a slower pace; by 23 percent in the agnculture sector and by 7 percent in constr~ction.~~Real wage trends in agnculture tumed a comer in 2000, after virtually stagnating in the 1990s. Figure 3.7: Real Wage Trends in Major Sectors (1991 normalized to 100) Wnufacturmgsectorwa Source: BBS Data and WorldBank Staff calculations 3.37. Real wage growth in the manufacturing sector exceeds the sector's labor productivity and output growth. While real manufacturing wages increased by 32 percent over 2000-2005, real manufacturing output per worker increased by no more than 20 percent.61 This disparity could arise from any combination of: a premium on certain skill sets not readily available; informational problems that prevent matching of jobs with available workers with the right skills, and; wage setting by collective bargaining. Other than collective bargaining, which doesn't seem to be prevalent in Bangladesh (Section VI), the first two forces are likely to be at play. There is some evidence of the skills premium increasing in the engineering industry. For instance, the ratio of the daily wage rate for a "skilled" fitter in the engineering sector to that of an "unslulled" fitter increased fiom 1.40 in 2000 to 1.72in 2005. 3.38. The disparity between real wage and labor productivity trends appears even more glaring for the agriculture sector. While real wages in the sector increased by 23 percent over the 2000-05 period, real output increased by just 13percent. In order for labor productivity changes alone to justify the real wage increases, labor force in agnculture would have had to fall by 10 percent over the five year period. While labor has been flowing out of agnculture, the pace is likely to have been slower than that. It is possible that part of the wage increase in agriculture was simply catching up with the productivity -- 59A nation-wide price index for the industrial sector has been used to convert the nominal indices into real indices for general, manufacturing and construction. To deflate nominal agriculture wages, price deflators for agriculture fiom the national accounts are used. 60The LFS data, on the other hand, show agriculture sector wages to have declined during 2000-2003, in large part because the 2003 survey was conducted during the lean agriculture season of October-November, while the earlier survey was spread over the year. 61This is assuming that the annual labor force growth in manufacturing was 3 percent, about the same as the overall labor force growth. The increase in output per worker in manufacturing over the 2000-05 period would be accordingly lower than 20 percent if the labor force growth in the sector was higher than 3 percent. improvements seen in the 1990s which were then accompanied by real wage stagnation. The adjustment would have occurred over the 2000-05 period because of the significant tightening of the labor market in manufacturing which trickled to other sectors, including agriculture. Two other important factors likely also contributed. One, certain high wage segments of agriculture - e.g. commercial poultry, high value crops etc. - deepened over the latter period, as shown in Chapter 2. Second, the forms of labor employment or features of contract are changing. Piece rate and contractjobs are beginning to substitute time rate or daily employment. Piece rate jobs are given for specific work and usually a group leader is involved in the negotiation. The employer does not supervise the laborers. This form reduces supervision cost and raises the intensity of work. Thus productivity and wage rates tend to benefit. Rural-urban wage differential 3.39. TheLFS data show significant rural-urban wage disparity in 2003; the weighted average rural wage rate was only 62per cent of urban wage rate (Annex Table 19): the divergence was 40percentfor males and 33 percent for females. The disparity between rural and urban wages stands even after accounting for the cost of living differences: cost of living in urban areas is estimated to be about 25 percent higher than in rural areas. Wage disparities in agriculture and services seem to be driving this (Figure 3.8). Urban wages for the agriculture sector were about 30 percent higher, while the wage gap in services varied between 22 percent (trade) to 65 percent (finance). In the manufacturing sector, urban wages are about 20 percent higher than rural wages, which mainly reflects the difference in cost of living. Figure3.8: Ratio Urban Wage Rate to Rural Wage Rate for Maior Economic Sectors (Percent) Source: BBS Data and WorldBank Staff calculations 3.40. The overall rural-urban gap remained virtually unchanged between 2000 and 2003, with offsetting changes across different sectors. The rural-urban wage divide became worse in the agriculture sector and improved dramatically in mining and also in the manufacturing and construction sectors between 2000 and 2003. The disparity didn't change much in most services sub-sectors, except personnel services where it became considerably worse. V. Determinants of Employment and Real Wages: An Analytical Approach Determinants of employmentlevels: cross-sectional variations 3.41. This section seeks to identify analytically the factors that affect weekly hours of employment in the major sectors. Key results from the regression reported in Annex Table 20 are: The size of landownership tends to increase employment levels. Here two offsetting effects would be at play. Bigger size of landownership can positively affect productivity and hence boost employment fiom the demand side. On the other hand, it can negatively affect employment fiom the supply side by reducing incentives to work. On balance, it appears that the positive demand side impact dominates. Workers training has a positive and significant effect on the hours of employment, likely reflecting an increased demand for their higher slull levels. Worker's education lowers the hours of employment. This is consistent with Table 3.2 which shows an positive relationship between unemployment and education. This likely reflects the higher reservation wages of the more educated and the substantial the gap that studies have found between the skills being demanded by the market and those being supplied by the system. Urban areas provide more hours of employment even after controlling for worker characteristics and sectors of employment. This likely reflects both demand and supply side factors. On the supply side, urban workers need to put in longer hours to cope with the higher cost of living and, for the large number of migrants, to be able to send remittance money back home to rural areas. On the demand side, productivity in urban areas is higher because of agglomeration economies and other benefits mentioned in Chapter 5. Among the various statuses of employment, hired employment is associated with more hours of employment than selfemployment - more so for the 'salaried employees' than wage laborers who are often hurt by seasonality of their work. Unpaid family workers and employers tend to put in fewer hours than the selfemployed. The manufacturing and construction sectors provide more hours of employment than agriculture. The trade and transport and communication sub-sectors provide more hours of employment that agriculture, but the rest of the services sector provides fewer hours. As would be expected, the relationship between hours of employment and age of the worker has an inverse U-Shape. Hours of employment increase with age up to a certain age and decline after that. This is also consistent with Figure 3.3b that shows a similar relationship with LFPR and age for male workers. Other worker and household characteristics also affect hours of employment. Males and heads of the family worker longer hours. Having more dependents in the households also increased the number of hours of employment. 3.42. The determinants of hours of employment are quite similar for the self-employed and for paid workers (Annex Tables 21 and 22). The coefficients on workers' education and training are not significant for the self-employed, while land ownership is not significant for either category. Determinantsof wage rates 3.43. Section IV showed rural-urban and sectoral differences in wage rates. It left out the micro characteristic such as human capital differences and household characteristics that could further help explain wage differences among workers. This section jointly assesses the role of all these factors that can affect wage earnings fiom paid employment. Regression results using 2002-2003 LFS data are presented in Annex Table 23, and show that: The public sector pays 32 percent higher wages, on average, compared to the private sector after controlling for all other factors, including skills and location. The premium for public sector is well recognized in ~an~ladesh,6~which results in rationing of these jobs to a few fortunate ones and also distorts wage signals in the rest of the economy. Urban wages are about 11per cent higher. This captures the urban-rural differencesin the cost of living as well as higher productivity in urban areas. Human capital - i.e., workers' training and years of schooling- tend to improve wages. This is only to be expected in an economyin which skill levels are still visibly quite low. Male wages are about 40 per cent higher than female wage rate, even after controlling for education and other worker, location and employment characteristics. Gender bias in remuneration is not unique to Bangladesh; even developed countries report evidence of it. Like the rest of the world, it is an important ongoing challenge for Bangladesh. Wage disparities perpetuate economicinefficiency-not to mention the social inequitythey perpetuate. Salaries of regular employees are about 20 percent higher than irregular employees and day laborers. This confirms that the workers in the former category are relatively more empowered to bargain their position. The higher remuneration also comes in addition to the better working conditions, includingsubstantiallyhigherjob security. Asset endowment -proxied here by size of land ownership-tends to improve wages. This likely has to do with the higher reservation wage of land owners. Those with bigger size land would be willing to work outside of it with better compensation. Why the employers would be willing to link pay with the size of landholding of the worker, everything else being the same, needs more analysis. Possibly, the enhanced social standing that comes with a larger land endowment improves the workers' bargainingpower as well. Differences in wages also exist among various sectors. Relative to the agriculture sector, the wage premium is 17percent in manufacturing, 32 percent in constructions,and anywhere from 2 percent to 30 percent in services. The relationship between workers age and wage rates has an inverse-U shape. Initially, wages increase with age up to a certain age, after which they decline with age. M. An Ovemew ofLaborLawsandRegulationsin~an~ladesh~~ 3.44. The Right of Association: Bangladesh's Constitution guarantees freedom of association, the right to join unions, and, with government approval, the right to form a union. With the exception of workers in the railway, postal, telegraph, and telephone sectors, civil servants are forbidden to join unions. The ban also applies to the military and police. Some public sector employee groups, such as teachers and nurses, although forbidden tojoin unions havejoined associationswhich perform functions. 3.45. The Industrial Relations Ordinance (IRO), 1969, provides that any worker or employer has the right to form a uniodassociation without previous authorization. But such a uniodassociation cannot function as a trade union without being registered under the law, which requires a 30 percent participation in the union before it can be registered, and a union may be dissolved if membership falls below this level. The KO's Committee of Experts on the Application of Conventions and Recommendations (CEACR) finds such requirements to be excessively restrictive and has requested that the government amend these. Additionally,the InternationalConfederation of Free Trade Unions (ICFTU) has expressed concerns that workers trying to establish a trade union are not protected by law, and as a result, are subjectto severe resistance fromemployers and government to prevent formation of the union. 62See, for example, Bangladesh Public Expenditure Review, World Bank, 2003. 63 Sources: U.S. (i) State Department CountryReport on Economic Policy and Trade Practices in Bangladesh, 1998. (ii) Fair Labor Association, 2004 Annual Report. (iii) Board of Investment, Bangladesh. 3.46. Bangladesh has ratified seven of the eight fundamental ILO conventions. In 1972 Bangladesh ratified both fundamental ILO Conventions concerning freedom of association: Freedom of Association and Protection of the Right to Organize (Convention No. 87, and Right to Organize and Collective Bargaining (Convention 98). However, when Bangladesh ratified the International Covenant on Economic, Social and Cultural Rights (ICESCR), it entered a reservation on articles 7 and 8, which guarantee the right of everyone to form trade unions and join the trade union of their choice. The reservation indicated that there would be some limitations placed on workers' freedom of association. 3.47. In 2002, there were about 6,300 registered trade unions, most with political affiliations,with less that 2 million worker members (less than 5 percent of the workforce). The relatively low level of union membership reflects the informal nature of vast majority of economic activity and the mentioned restrictions that apply to the formal sector. 3.48. The Rinht to Bargain Collectively: Collectivebargaining by workers is legal under the condition that their unions are legally registered as collective bargaining agents. In case of multiple registered unions in an establishment, a collective bargaining agent is determined by the Registrar of Trade Union - a position given to the Director of Labor of the government - through a secret ballot for a term of two years. Collective bargaining occurs occasionally in large private enterprises but is generally not seen in small private enterprises. 3.49. Right to Strike: Strikes are common in Bangladesh and are recognized in the 1969 IRO as a legitimate avenue for addressing unresolved grievances. The IRO permits the government to bar strikes for three months in any sector deemed "essential." These include utilities (electricity, gas, oil and water etc.), hospital and ambulance services, fire brigade, railways and the national airline Birnan, and ports. Mechanisms for conciliation, arbitrationand labor court disputeresolution were also establishedunder the 1969IRO. 3.50. The ILO Committee of Experts has objected to the following provisions of the IRO on the grounds that these restrict workers' right to defend their economic intereststhrough strikes: the need for three-quarters of the members of a workers organization to consent to a strike; the government's power to prohibit a strike if it lasts more than 30 days or to prohibit a strike at any time if it is considered prejudicial to the national interest; and penalties (which include imprisonment) that may be imposed if workers participate in an industrial action that is deemed by the governmentto be unlawful. 3.51. Prohibition of Forced Labor: The constitution prohibits forced or compulsory labor. The Factories Act and the Shops and EstablishmentsAct, both passed in 1965, set up inspection mechanisms to guard against forced labor, but resources for enforcement are scarce. Nevertheless, the practice of forced labor is believed to be quite limited. 3.52. Minimum Age for Emvlovment of Children: Bangladeshhas laws that prohibit labor by children. The Factories Act bars children under the age of 14from working in factories. The enforcement of these rules is quite inadequate though and the application of the rules to informal sector activities such as agriculture and domestic help -where majority of child labor is found -is not clear. As per the 2000 LFS data, over 5 million, or 20 percent of the population in the 0-14 age group, was working. 3.53. In July 1995, Bangladesh garment exporters signed an MOU that has virtually eliminated child labor in the garment export sector. Under the MOU, schools and a stipend program were established for displaced child workers. By November 1998, hundreds of schools serving thousands of former child workers were in operation. A system of fines and possible suspension of import/export privileges exists, and a monitoring system has been set up by the ILO. 3.54. Acceptable Conditions of Work: Regulations regarding minimum wages, hours of work and occupational safety and health are not strictly enforced. The legal minimum wage varies depending on occupation and industry. It is generally not enforced in the private sector. The garment sector has witnessed prolonged worker agitation over dissatisfaction over minimum wage rates which have not changes for several years. The impasse carries on unresolved, highlighting the significant economic cost of not having in place a functioning mechanism to resolve labor disputes in the formal private sector. In the public sector, wages and fringe benefits of the workers are determined on the recommendation of the National Pay Commission established fi-omtime to time. Sometimes private industries follow the public sector wages & salary structure for their workers. 3.55. Leave & holidays of the workers & employees are regulated by the Factories Act, 1965 and shops Establishment Act, 1965. The law sets a standard 48-hour workweek with one mandated day off. A 60- hour workweek, inclusive of a maximum 12 hours of overtime, is allowed. The Factories Act of 1965 nominally sets occupational health and safety standards. The law is comprehensive but appears to be largely ignored by many employers. Freedom of Associationin Export Processing Zones in Bangladesh 3.56. In 1980,the Bangladesh Export Processing Zones Authority Act 10 was enacted. It provided for the establishment of the Bangladesh Export Processing Zones Authority, (BEPZA), which is the official arm of the government responsible for the creation, development, operation, management, and control of export processing zones (EPZs). Under the BEPZA Act the government may exempt an EPZ from as many as 16 laws, including the Industrial Relations Ordinance (IRO). In 1986, the government declared, in accordance with the BEPZA Act, that the IRO was not applicable in the EPZs. The declaration effectively suspendedthe rights of workers in EPZs to freedom of association and collective bargaining. 3.57. Mostly under pressure from the U.S., in 2004 the Parliament of Bangladesh passed a bill granting limited workers' association rights in the country's export processing zones (EPZs) effective from November 1,2006. Under the 2004 Bill, trade unions of a particular EPZ can form a federation, but there cannot be more than one federation in an EPZ. The federations will not be allowed to form a single national body orjoin any national trade union, political party or its labor front. VII. Challenges Ahead 3.58. A rapidly growing labor force presents a critical development challenge for Bangladesh. Its labor force is projected to increase by 10 to 20 million over the next decade - 10 million if the female LFPR does not change and 20 million if female LFPR increases to 40 percent. Creating productive employment opportunities will clearly be an imperative for poverty reduction and even sustained economic growth in Bangladesh, as rightly recognized by the Government's Poverty Reduction Strategy (PRS). Labor market regulations in Bangladesh do not appear that encumbering, especiallyrelative to India, as is also borne out by various investment climate survey results which show that hiringlfiring rigidities are not among the main concerns of employers. More effective enforcement of the existing labor market policies is needed, nonetheless, especially to protect worker rights. Instrumental for meeting the government's employment generation objective will be sustained productivity and output growth across all major economic sectors (an agenda covered in the rest of this report); deeper and more focused government attention to enhancing worker skills, particularly by strengthening the higher education and VET systems and making them more responsive market needs; improved urban management to facilitate rural-urban migration; and, continued emphasis on gender empowerment to allow greater mobility of female workers to more productive employment opportunities. Chapter 4: Trade Liberalizationand Export Competitiveness I. Introduction 4.1 Starting in 1991, Bangladesh marked a clear departure from the highly protectionist, inward- looking import substitution policies of the earlier decades, with tangible benefits to GDP growth. Considerable, albeit uneven, progress has since been made on liberalizing the trade and exchange regimes. Trade tariffs have been significantly reduced and rationalized, with the unweighted average customs duty (CD) falling from 100percent in FY85 and 74 percent in FY92 to 15percent in FY07, and the top CD rate from 350 percent in FY90 to 25 percent. Trade-related quantitative restrictions (QRs), which covered 21 percent of HS 4-digit tariff lines in FY90, have been eliminated, and import licenses are no longer needed. Foreign exchange liberalization has included unification of exchange rates, removal of most foreign exchange restrictions for current account transactions, and adoption of a floating exchange rate policy. 4.2 Despite good progress, a substantial agenda for deepening trade liberalization and strengthening export competitiveness remains. Although the average CD rate has been steadily reduced by cutting the general maximum rate, a rampant increase in the use of other protective levies (para-tariffs) in the last decade has offset much of that. As a result, the average nominal protection in Bangladesh is the highest in the region and among the highest in the world. Furthermore, trade facilitation procedures are cumbersome and infrastructure bottlenecks to exporting are severe. The resulting anti-export bias has leveled off at around 20 percent since the late 1990s,which works against the emergence of new exports and expansion of the export activities to non-enclave areas. Unsurprisingly, Bangladesh's export orientation, although much higher than in 1990, compares unfavorably with other Asian economies. Moreover, the export base is heavily concentrated in the readymade garment (RMG) sector - which accounts for 75 percent of merchandise exports -where the extent of the anti-exportbias is much lower. 4.3 The potential net benefits fiom further opening up the trade regime are high for -certainly higher than for the larger emerging economies such as China and India. In fact, even these larger economies are now moving decisively down the trade liberalization track to reap its benefits. The impressive growth performance of many successful East Asian economies was shaped in large part by their ability to integrate into the global supply chains at early stages of development. At the heart of their increased integration was lowering of trade bamers and anti-export bias and improving trade facilitation. Gradually, with economic development, strengthened domestic competitiveness, and improved labor skills these economies moved up the quality ladder toward higher-end production chains, achieving unprecedented GDP growth in the process. A broad strategy along those lines holds tremendous potential for ~ a n ~ l a d e s This h . ~ ~chapter presents an overview of the trade policy trends in Bangladesh and makes some policy recommendations that would help strengthen the country's global competitiveness. The trade facilitation aspect of competitiveness is covered in Chapter 9. 11. Trends in Trade Policy and the Anti-Export Bias Overview of Trade Policy Trends Since Early 1990s 4.4 Considerable, albeit uneven, progress has been made on trade and exchange liberalization since 1991. Reform measures have included tariff cuts and rationalization, removal of trade-related QRs, adoption of a unified exchange rate system, switching from a fixed to a more flexible exchange rate This combination of trade policy reforms and strengthening domestic competitiveness also appears to be the best policy response to the challenges and risks emanating from the phasing out of the ATC quotas in January 2005. regime, and current account convertibility. These have been accompaniedby complementary measuresto liberalize the domestic economy and improve market orientation: contributing to the improved growth since the early 1990s. 4.5 More substantial progress on trade liberalization measures was made during thefirst half of the 1990s. During this period, import tariffs were cut significantly through a 'tops down' approach of reducing the maximum CD rate while also reducing the number of tariff slabs. As a result, the (unweighted) average protection rate fell from 73.6 percent in FY92 to 32.4 percent in FY96 and the dispersion of tariffs declined by over 60 percent (Table 4.1). The number of QRs fell significantlyin the 1993-95Import Policy Order (PO) relative to the 1991-93 P O (Table 4.2) Table4.1: Average ProtectionRate in Bangladesh (MFNRates, includingconcessionalrates in %) " 1991192 1995196 2003104 200415 2005106 2006107 Unweighted average Customs Duty (CDl- MFN rates only 70.6 28.7 18.8 16.3 15.5 14.9 Industrial products1a 69.7 28.4 18.0 15.6 14.7 14.0 Agriculturalproductsm 76.6 30.1 22.6 19.6 19.4 19.1 Unweiehted total protection rate-MFN rates only 73.6 32.0 29.1 26.5 26.5 24.3 Industrial products 73.2 31.9 26.8 25.4 25.3 22.4 Agriculturalproducts 76.6 32.4 39.8 32.1 32.4 31.3 Standarddeviation of total protection 41.9 15.9 25.9 25.5 27.1 22.6 {based on MFN tariff lines) Tariff lines subiect to protectiveSD and VAT SupplementaryDuty .......................... 49 13 691 1066 1031 1043 VAT .......................................... 426 373 727 582 600 601 Unweighted avg. CD-MFN rates, adiusted for tariff exemptionslconcessions 57.6 22.6 15.6 13.5 12.4 11.8 Industrial products 55.9 21.5 14.6 12.6 11.4 10.8 Agriculturalproducts 73.5 29.4 22.2 19.3 18.9 18.7 Unweightedtotal protection rate-MFN rates, adiusted for tariff exemptionslconcessions 60.2 25.1 23.7 21.4 20.7 18.9 Industrialproducts 58.7 24.0 21.2 19.9 19.1 17.3 Agriculturalproducts 73.6 31.6 38.9 31.4 31.6 30.5 Percentageand the number of tariff lines with 93.3 86.1 72.5 70.1 70.8 69.4 internationaltariff peaks (total protection rate >IS%) (6233) (5810) (4983) (4699) (4746) (4649) I 1: Averages reported here are based on 8digit 6877tariff lines and the MFN tariffs, and do not reflect 'preferential' tariffs. I a: Harmonized System(HS) Chapters 25-97; Ib: HS Chapters 01-24. Ic: 'Total protection' incorporates,in addition to customsduty, the protective effectiveof the para-tariffs. Source: TradePolicies in South Asia: An Overview, WorldBank, 2004; Bangladesh's Import Policies, World Bank, 2005. 4.6 The trade liberalization drive slowed down considerably during the FY97-99period, with even some instances of reversal. The average rate of nominal protection rose slightly over the period and the number of HS 4-digit items subject to QRs increased from 111 in the 1993-95 P O to 122 in the 1997-02 PO. Resistance by domestic protectionist lobbies, the heavy reliance on customsrevenues, and concerns about the slow progress in addressing domestic business environment weaknessesplayed a key role in the slow down. Many openly womed that Bangladesh was perhaps opening up too much, too fast. These fears turned out to be unfounded, however.65 Bangladesh had started the trade policy reforms from prohibitively high protection levels to begin with, and a significant portion of the initial tariff cuts essentially amounted to eliminating 'water-in-the-tariff (i.e., tariff redundancy) in the case of final products. More importantly,trade liberalizationwas accompaniedby sustained strongperformance of the export sector and the economy in general. The BOP was also not overwhelmed by a surge in imports 65This point is forcefully made by Bangladesh: TradeLiberalization: Its Pace and Impact, World Bank, 1999. during periods of import liberalization, as many had predicted: in fact, export growth outpaced import growth duringthese periods. 4.7 Trade liberalization was partially revived during FY00-FY02, and picked up pace in FY03. The average CD was cut from 26.6 percent in FY99 to 21 percent in FY02, and further to 14.9 percent in FY07. The decline in average nominal protection was more modest, however, because of increased use of other protective import levies, or para-tariffs (see below) -nominal protection fell fiom 32.4 percent in FY99 to 24.3 percent in FY07. In the meanwhile,the number of QRs has been lowered to 15(Table 4.2), of which only three are protection-related (poultry, eggs, and salt), and on these GOB has obtained waivers from WTO until 2009. More significantly, the long-standing import bans on many textile products have also been removed, strengthening the competitiveness of the critical garment manufacturingsector. Table4.2: Evolution of ImportRestrictions 1991-2006 IPO IPO IPO IPO IPO 1991-93 1993-95 1995-97 1997-02 2003-06 (JULYO~) Number of items in the control list at the 193 111 120 122 48 HS 4-digit level (15.6%) (9.0%) (9.7%) (9.8%) (1.4%) Number of trade-related items in the 79 19 27 28 3 controllist at the HS 4-digit level (6.4%) (1.5%) (2.2%) (2.2%) (0.3%) Source: WTO Trade Policy Review, Bangladesh 2000; IPOs various years. 4.8 A pattern of increasing reliance on para-tariffs for protection has emerged over the last decade, to the detriment of trade openness. Sincethe mid-1990s, the Governmenthas increasinglyrelied on the para-tariffs for protection as well as revenue purposes, partially offsetting cuts in C D S . ~Four ~ types of para-tariffs have been put to use: InfrastructureDevelopment Surcharge,Regulatory Duty, Value Added Tax (VAT), and SupplementaryDuty (sD).~' The RegulatoryDuty, introduced in FYO1, was later removed in FY05. The para-tariffs now cover about 20 percent of tariff lines and account for just under 40 percent of total nominal protection provided by the statutory import duties, as opposed to 10percent in ~ ~ 9 6These' developmentshave led to an increase in the dispersionof total tariff protection, from 15.9 . ~ percent in FY96 to 22.6 percent in FY07. Moreover, introducingprotection through the back door - i.e., through increased use of para-tariffs - increases the scope for lobbying pressures by protectionist groups and for corruption. 4.9 Especially worrisome is the recent widespread use of SDs, which have become a key instrument of trade protection. SDs now cover 1,043 tariff lines, up from 356 tariff lines in FY03. The increased reliance on their use was most visible in the FY05 budget. Then, in a far-reaching move, the Government reduced the maximum CD rate from 30 percent to 25 percent, moved from a four to a three non-zero tariff slab system, and eliminated the Regulatory Duty. But it also fkrther increased the number of tariff lines subject to SDs, from 691 in FY04 to 1,066 (Table 4.1), offsetting much of the impact of the other liberalizationmeasures. 66For a detailed discussion on the protective application of the para-tariffs, see: Trade Policies in South Asia: An Overview, Chapter 3; and Bangladesh S Import Policies, World Bank, 2005. 67In principle, both the SDs and VAT are supposed to be protection-neutraltaxes. In practice, however, the SDs are levied mostly on only imported products, and when applied on domestic goods, the rates are generally lower. Furthermore, because the SDs are applied on the landed value plus CD, it has a cascading effect, with protection rates rising further with higher CD rates. Similarly, VAT has been used to provide extra protection to certain import ' competing industries (e.g., imports of textiles are subject to the general VAT rate of 15 percent, but domestically roduced textiles are subjectto a special lower VAT rate of 2.5 percent). This was despite the elimination of the License Fee in FY03, an across-the-board import tax of 2.5% of the assessablevalue, and also the removal of the Regulatory Duty (of varying rates) in FY05. 4.10 There is signijicant tariff escalation in the import duty structure, with the CD rates rising according to the stages of processing. This means that the effective rates of protection for import substituting final products are considerably higher than the nominal tariff protection on these products. This escalation is magnified by the imposition of the SDs whose duty-base includes landed costs plus CD. For example, given the general maximum tariff rate of 25 percent in FY05, with an SD rate of 25 percent, total nominal protection for tableware products reached 84 percent.69 The calculations for major commodity categories presented in Table 4.3 show tariff escalation as well as the cascading impacts of para-tariffs since FY92. Moreover, because of a plethora of end-user specific CD concessions, tariff policies have discriminated against the domestic production of certain products that were not favored, such as intermediate material and components.70 Table 4.3: Bangladesh: Tariff Escalationand CascadingImpacts of Para-Tariffs on Total Nominal Protection UnweightedAverage Rates (based on MFN tariffs only, %) Description CD CD+DS+RD PSD I' P-VAT I' Total Nominal Protectionfrom FY +LF I' Protection para-tariffs(%) 06-07 Basic raw materials 9.3 12.8 0.9 0.0 13.7 31.7 06-07 Intermediate inputs 12.3 16.1 1.1 -0.1 17.2 28.2 06-07 Capital goods 9.2 13.1 1.7 0.1 14.9 38.7 06-07 Final consumer goods 20.2 24.1 8.5 3.3 35.8 43.6 06-07 Overall 14.9 18.7 4 3 1 3 24.3 38.8 03-04 Basic raw materials 12.1 15.9 0.2 0.2 16.2 25.6 03-04 Intermediate inputs 16.9 20.8 1.4 0.3 22.5 25.0 03-04 Capital goods 12.0 15.9 3.1 0.2 19.2 37.8 03-04 Final consumer goods 24.0 30.5 6.3 3.3 40.2 40.3 03-04 Overall 18.8 23.8 3.7 1.6 29.1 35.4 95-96 Basic raw materials 14.9 17.4 -0.1 0.1 17.4 14.4 95-96 Intermediate inputs 26.3 28.8 0.0 0.9 29.6 11.3 95-96 Capital goods 20.6 23.1 -0.2 0.2 23.0 10.5 95-96 Final consumer goods 35.9 38.4 -0.3 1.4 39.4 9.1 95-96 Overall 28.7 31.2 -0.2 0.9 31.9 10.1 91-92 Basic raw materials 91-92 Intermediate inputs 91-92 Capital goods 91-92 Final consumer goods 91-92 Overall I* CD: Customs Duty; DS: (Infrastructure)DevelopmentSurcharge; LF: LicenseFee; RD: Regulatory Duty; SD: SupplementaryDuty; VAT: Value-Added Tax. P-SD: Protectivecomponent of SD; P-VAT: Protectivecomponent of VAT. Source:Staff calculations. 69For more examples of product specific tariff escalation by stages of processing, see: TradePolicies in South Asia: An Overview, page 64, Table 3.17. 70These calculations are based on the protection levels made 'available' by import duties. In practice, statutory tariffs may not reflect the actual price differences between domestic and international prices due to illegal border trade and smuggling through official channels. However, it is important to stress that such informal trade is induced, in part, by high protection rates, and it is the structure of protection shaped and made available by the statutory duties that is going to affect medium- and longer-termproduction decisions. Anti-Export Bias Remains High 4.11 Anti-export bias, measured here as the ratio of (average) effective exchange rate for imports (EERrn) to that of exports (EERx), is often used as an indicator of the trade regime's anti-export bias.71 The higher is the ratio above 1.O, the higher the bias against export a~tivities,~'and the more damaging it is for export competitiveness. Box 4.1 discusses the various channels through which trade protection impacts the anti-exportbias. Box 4.1: Trade Protection: Its Impact on Anti-Export Bias and Export Competitiveness Enhancing Bangladesh's export competitiveness depends crucially on reducing the anti-export bias of the trade regime, which, in turn, is shaped by the level of trade protection. The tariff structure impacts the anti- export bias through severalchannels: Duties levied on imports of final goods raise their domestic relative prices, thereby increasing the profitability of import substitutes relative to exports. This diverts resources towards inefficient production for the domestic market, away from the production of exports. For example, the very high protection (nearly 100 percent) afforded to the ceramics industry e.g., tableware - reflected in much higher domestic prices of ceramics products relative to their export prices - appears to be discouraging production for exports. An escalatingtariff structure raises the effectiveprotection for an import substituteabove the nominal protection that the same import substitute receives from import duties and other protection. This means that the value-added (processingmargins) involved in production for the domesticmarket will exceed the value-added that would have existed in the absence of any protection, by proportionately more than the nominal protectionof the final product. This M e r increasesthe anti-exportbias. With import demand being curtailed under high protection, import-related (ex-ante) demand for foreign exchange is also curtailed, enabling the country to maintain a lower exchange rate (i.e., a lower domestic currency price for foreign currency) than otherwise. This then translates into lower domesticcurrency proceeds for export. Exporters sell in competitive world markets and cannot pass on higher production costs to buyers. Thus, import duties paid on imported inputs increase their production costs and cuts into their profit margins and make them less competitiveglobally. Even if there is a dutyltax rebate system, these are generally very inefficient and slow, and even the delayed collection of rebates have associated costs (e.g., bribes). Enclave arrangements, such as the bonded-warehouse scheme used by the RMG sector and export processing zones (EPZ), provide speedy duty-free access to imported inputs, but they serve only specificllimited activities and, as such, they are not substitute for extended periods for broader and deepertrade liberalization. 4.12 Reflecting the patterns of trade liberalization, the anti-export biasfell significantly in thefirst half of the 1990s, and has leveled off at a relatively high rate since the late 1990s (Table 4.4). This holds for both the average import effective exchange rate based on the MFN rates only (column 9) as well as with the import EERs that also reflect end-user tariff concessions/exemptions (column 10). The lack progress in the latter period was on account of declining amount of direct cash support and other forms of explicit subsidies (such as provision of subsidized credit) for exporters, which offset the reductions in average nominal protection. The increased use of the para-tariffs has also contributed to the high levels of 71EERs for imports refer to nominal exchange rates adjusted for (protective) import levies and any scarcitypremium that exchange controls may generate. The EERm here indicates the domestic currency cost of one US$ worth of imports. For exports, EERx represents the exchange rate after adjusting for existing export promotion schemes (direct export cash subsidies, subsidized export credits etc.). Thus, EERx represents domestic currency equivalent of proceeds fromexports worth one US$. 72A ratio of 1.0would imply a trade policy that is neutral towards exporting and import substitutingproduction. anti-export bias by holding back the reduction in nominal protection. With the extent of average bias remaining at about 20 percent, the structure of incentives created by the trade policy still favors the production of import substitutes and constitutes a significant bamer to the emergence of new export areas and to the expansion of exports that are not benefiting fiom enclave arrangements cited below. Table 4.4: Estimates of Anti-Export Bias Based on Effective Exchange Rates for Imports and Exports Fiscal Imports Average total Averagetota Nominal EERm E E R ~ EERx ~ ~ Anti-ExportBias Year (in billion protection rate protect. rate Exchange rate tk) (%) " (%)" (TWS$) EERm/ IEERmll correspondingEERm is given in column 7. The latter also reflectsduty-free access to imported inputs provided to 100percent 4.13 Bangladesh has managed to insulate certain activities- most notably the RMG sector -from the anti-export bias, through the bonded-warehouse and Export Processing Zones (EPZs). Furthermore, exporters without access to the duty-free bonded warehouse system or EPZs (because they are not 100percent exporters) and required to pay import duties or domestic taxes on imported inputs, can benefit from the existing duty drawback system managed by the Duty Exemption and Drawback Office (DEDO). While such enclave type arrangements provide cheaper access to certain types of exporters, they do not offer shelter from the broader disincentives caused by high protection listed in Box 4.1. 4.14 Moreover, the DEDO suffersfrom weak governance and inefficiencies. Exporters experience significant delays in payments of tariffltax rebates, despite various attempts to strengthen the DEDO through staff training, new technical staff, and strengthenedreimbursement procedures. This is partly due to the complexity of the existing import duty regime and a very lengthy duty drawback process involving the determination of input-output coefficients and the submission of 18 documents by direct exporters (even more documents are required in the case of indirect exporters). It takes a minimum of six months for drawbacks to bepaid; 50percent of applications are not disposed of in twoyears. 4.15 Securing both the continued success of the RMG sector and export diversification will require sustained efforts to reduce the anti-export bias through further trade reforms, along with measures to address the key behind-the-border investment environmentbottlenecks discussed elsewhere in this report. BangladeshLags the Rest of the World on Trade Openness 4.16 Whiletrade liberalization occurred injits and starts in Bangladesh, stronger and more decisive commitment to trade liberalization was seen in majority of the rest of the world. As a result, Bangladesh lags behind on most measures of trade openness (Table 4.5). Even after the reduction in nominal protection in the FY07 budget, Bangladeshhas the highest level of trade protection in the region, which itself is the most trade restrictive region in the world. Other South Asian countries, particularly India and Pakistan, have cut their protection levels far more aggressively in recent years, while Sri Lanka started its trade liberalization in the late 1970s and has continued to maintain much lower protection levels (except for the agricultural products). In fact, in 2004, only 4 countries in the world had higher average nominal protection than Bangladesh, and it is unlikely Bangladesh's rank has improved since. Trade liberalizationdid lead to an increase in Bangladesh's trade-to-GDP ratio - from 20 percent in 1990 to 38 percent in 2005. However, since trade liberalizationwas a common phenomenon across the globe over the period, Bangladesh's global rank on the measure did not change by much. Among the 140 countries for which comparable data are available for 1990 and 2004, Bangladesh ranked 1 3 5in ~ 1990 and slipped to 137' by 2004 in terms of the trade to GDP ratio.73 Table4.5: Average MFN tariffs in South Asia and other select developingcountries All products (134 countries) Agriculture (134 countries) Averagetariff (%) Rank Year Average tariff (%) Rank Year Morocco 33.4 1 2002 Morocco 53.6 1 2002 Tunisia 30.2 3 2002 Turkey Bangladesh 26.5 5 2004-05 Tunisia Iran 23.9 7 2002 Korea Nigeria 23.4 8 2002 India India 22.2 10 2004-05 Iran Pakistan 18.5 19 2002-03 Bangladesh Egypt 18.4 20 2002 Sri Lanka Nepal 18.0 22 2003-04 Mexico Mexico 16.2 30 2002 Nigeria Vietnam 15.0 32 2001 Pakistan Ghana 14.7 34 2000 Ghana Thailand 14.7 35 2002 Vietnam Sri Lanka 13.4 42 2003-04 Nepal Turkey 12.6 47 2001 Egypt Korea 12.6 48 2002 China Brazil 12.3 51 2002 Thailand China 12.3 52 2002 Colombia Argentina 11.8 62 2001 Argentina Colombia 11.7 63 2003 Brazil Malaysia 8.8 86 2002 Philippines Indonesia 7.2 99 2002 S. Africa Chile 7.0 101 2002 Indonesia S. Africa 6.4 106 2001 Chile Philippines 5.1 120 2003 1 Malaysia 3.0 129 2002 Median 11.2 1 Median 15.1 Mean 11.7 I Mean 16.7 ;ource: WorldBank (2004): Trade Policies in South Asia: An Overview,page 35. 111. Trade Liberalizationin Bangladesh:Impact and Remaining Challenges Growth has Benefited from Trade Liberalization 4.17 The intended objectives of Bangladesh's tradepolicy reforms initiated in the early 1990s were to foster trade integration, enhance domestic competition, and align domestic relative prices with 73Source: World Development Indicators, World Bank. internationalprices. The expectationwas that these developmentswould promote efficiency in resource use and encourage technological progress and difhsion and thus generate dynamic gains and higher productivity growth. This reasoning drew on the widespread agreement that in the long-run economies with liberal trade policies and greater openness show stronger economic growth and overall development performance, based on supporting evidence from numerous cross-country studies and country case studies.74 Has trade liberalization in Bangladesh brought with it the intended consequences of growth and eficiency gains? 4.18 Trade liberalization and the resulting increase in export orientation in Bangladesh since the early 1990s has certainly coincided with and very likely contributed to faster GDP growth. This is supported by evidence at the micro and macro levels. Chapter 1 of this report shows that export orientation and productivity across manufacturing f m s in Bangladesh are positively correlated. A similar conclusion was reached by a previous World Bank study.75 At the macro level, too, growth performance, especially of manufacturing sector and exports, has improved considerably since 1991, benefiting from the trade reforms.76 4.19 A large part of the growth response of exports camefrom the RMG sector, which, in addition topreferential access to developed country markets, benefitedfrom far deeper trade liberalization than did the rest of the economy. The Government moved early to avail of the opportunitiespresented by the 'reserved' import markets under the Multi Fiber Arrangement (MFA) and more recently under the Agreement on Textiles and Clothing (ATC) until its complete removal in January 2005 and by the Generalized System of Preferences (GSP) and 'Everything But Arms' programs in the European Union markets. It expanded the use of 'special bonded warehouses' to facilitate quick duty-free access to imported inputs for the 100percent exporters, offered back-to-back letters of credit, and established EPZs to bypass some of the business environment weaknesses besetting the rest of the economy. Authorities also substantially lowered tariffs on imports of raw materials and capital equipment for most finished products. Several other industries, that had the potential to show a similar growth response, were held back by the slowpace of trade liberalization. 4.20 Thepositive experience of the RMG sector is testament to the benefits of reducing protection levels while improving business environment and trade facilitation. It reflects the capacity of the Bangladeshi private sector to respond to the new opportunities and the significant positive externalities created through the learning-by-doing process, as evidenced by the emergence of a new generation of entrepreneurs in Bangladesh. Important ChallengesRemain Because of the Uneven Nature of Trade Liberalization 4.21 The impressive success of the RMG sector is yet to be replicated in other industries, and Bangladesh's export base remains narrow. As noted, in large part this is because trade liberalization has been much slower in reaching the non-RMG sectors. If anything, Bangladesh's RMG experience demonstrates that relying on enclave-type arrangements to facilitate export growth in a specific economic activity, while postponing broader and deeper trade liberalization, can even backfire in some ways by creating significant concentration of the export base. Today, knitwear and woven-based RMG exports 74Dollar and Kraay (2002) and (2004), Michaely, Papageorgiou, and Choksi (1991), Winters, McCulloch, and McKay (2004). There are also studies reflecting skepticism over the findings of cross-country studies on the grounds of inappropriate measurement of 'trade policy-induced openness', weaknesses in the methodologies used, and failure to establish the direction of causality: Rodriguez and Rodrik (2000). However, the latter is in turn criticized for ignoring the findingsof the country case studies: Srinivisanand Bhagwati (1999). 75The Bangladesh- TradeLiberalization:Its Pace and Impacts, World Bank, 1999,ReportNo. 19591-BD. 76Ahrned and Sattar (2003)make the case that the improved growthperformancesince the early 1990swas in significantpart due to the trade reforms. account for 75 percent of Bangladesh's merchandise exports and directly employ 1.8 million workers (90 percent being women), and indirectly another 0.8 million workers in the accessory industries. In this post ATC era, with the abolishment of the quota system, such heavy economic concentration presents significantvulnerabilities that need to be recognized. 4.22 In order to diversifl its trade base and improve export competitiveness in general, Bangladesh needs to develop and implement a well-thought-out sequence of trade liberalization measures. The country has to accept the reality of increasing global economic integration that is likely to persist, intensifying competitive pressures among nations. Beyond that, global competition for Bangladesh's main export, RMGs, itself has stiffened with the phase-out of the textiles and clothing export quotas. Low wages notwithstanding, competitiveness of Bangladeshi exporters continues to be undercut by low productivity,higher input costs, perceptions of poor product quality, inefficienciesin trade facilitationand long lead times, and weaknesses in the domestic business environment; areas where many of Bangladesh's main competitors for RMG exports (e.g., China, India, Sri Lanka, Vietnam) are generally doing better. In order to maintain its position as a leading RMG exporter and also to diversify its export base, Bangladesh would need to strengthen its export competitivenessby addressing each of these critical areas of concern. IV. Towards Growth-SupportingTradePolicy Strategy 4.23 Over the last two decades, Bangladesh has followed a unilateral approach as the principal avenue to liberalize its trade policy, while remaining active in the ongoing WTO multilateral trade negotiations,particularly as a leader in the Least Developed Country (LDC) coalition under the Doha Development Round. However, in the area of 'offers' in agricultural and non-agricultural trade, Bangladeshhas opted for a strategyof keeping the coverage of its 'tariff bindings' very low (at 13percent of all tariff lines so far). Furthermore, the bound rates have also been kept significantly above the applied rates for both agricultural as well as for manufactured products.77 In any case, given Bangladesh's prohibitively high protection rates in the early 1990s, trade liberalization through unilateral actions (of course, in a manner that is consistent with the multilateral system of the GATT principles) and not waiting for the multilateral trade negotiations as a key instrument of Bangladesh's own tariff rationalizationeffortsseems to have been the right strategy. 4.24 Bangladesh and the other South Asian countries have also been actively involved in the negotiations of various regional 'preferential trade agreements' (PTAs) and in discussions ofpotential bilateralfree trade agreements (FTAs). Quite apart from general opening up through unilateral trade liberalization particularly since the early 1990s, South Asian countries view increased cooperation and trade among themselves as a key objective. This was reflected in the signing by the member nations of the South Asian Association of Regional Cooperation (SAARC) - Bangladesh, Bhutan, India, The Maldives, Nepal, Pakistan, and Sri Lanka - of the South Asian Preferential Trade Area (SAPTA) Agreement in 1993, which became operational in December 1995. The actual exchange of preferences has been quite limited, partly due to the hostilities between India and Pakistan, and the impact of this initiativeon Bangladesh trade liberalizationand trade opennesshas been minimal. 4.25 More recently, the worldwide proliferation of PTAs has led to a change in thinking in the region, especially in India but also in other South Asian countries, which have all begun to negotiate a series of PTAs of their own. As in other regions, these efforts particularly intensified in the aftermath of the failed September 2003 Cancun Ministerial meet. The South Asian Free Trade Area (SAFTA) Agreement was signed in January 2004, with the ultimate objective of turning South Asia into a full- 77For details, see: TradePolicies in South Asia: An Overview,World Bank, 2004, Report No. 29949. In FY04, the simple average of bound tariff rates was 188 percent - 197 percent for agricultural products and 50 percent for manufactured products. fledged FTA with the internal liberalization beginning in January 2006. This agreement has come in the wake of a bilateral FTA agreement between India and Sri Lanka in 1998 that became operational on March 1,2000. At the same, there is an ongoing discussion between India and Bangladesh on a possible bilateral FTA. Outside the regon, Bangladesh is discussing details of an FTA with ~ a l a ~ s i a . ~ ~ 4.26 Thepotential costs of PTAs/FTAs need to be carefully evaluated. Given that South Asia is the most protected regions of the world, Bangladesh faces considerable risks of adverse 'trade diversion' effects from regional PTA/FTA agreements. This is because such (multilateral and bilateral) preferential trading arrangements may lead to shifting of the source of imports away from least cost/most efficient third countries to higher cost members countries. In addition, a loss of customs revenues could also result.79 It is therefore critical that Bangladesh continues to reduce the high protection levels with unilateral trade policy reforms by reducing the average level and dispersion of import tariffs, with a particular effort to dismantle the existing para-tariffs. This strategy will also enable Bangladesh to better manage its 'regional integration' objectives by helping to minimize the adverse 'trade diversion' effects. 4.27 Specific recommendations are listed below, first with respect to the unilateral trade liberalization strategy, and then on the key elements of a strategy toward potential FTAs, assuming that the Government is determined to remain a member of the SAFTA and finalize the ongoing bilateral discussions with a view to signingFTA agreements. 4.28 Recommendationsfor thefuture unilateral trade liberalizationprogram: Establish a low and uniform tariff rate in not so distantfuture - the key ultimate objective of the future trade reforms. And these interim reform steps will need to include further simplification of the import tax regime, and reduction in the dispersion and average level of nominal (and thus effective) protection, preferably through a pre-announced medium- and long- term schedule of tariff reductions (as done recently by India). Unifl allpara-tariffs and merge these with the CD, with the result of having one tariff rate for each tariff line (in one-to-two years); Following the practice of the last decade, a 'tops down' tariff reduction approach could be maintained. Accordingly, gradually reduce the maximum tariff rate (afCer the merging of all para-tariffs with CDs) to 20percent over the next two-threeyears,followed by a new maximum tariff rate of 15 percent by, say, 2010. This would, of course, also require parallel progress in expanding the base of VAT and direct taxes, elimination of the numerous tax exemptions, and further strengthening of tax and customs administration; Eliminate all end-user tariff exemptiondconcessional tariffs; Discontinueprotective use of VATby ensuring that it is levied both on domestic production and imports of the same products (in one year); Eliminate all the remaining quantitative restrictionshuns that areforprotectivepurposes, such as those applying on imports ofpoultry, cartons, and salt. Replace them with appropriate tariffs (in one year); To maximize benefits from the EU's GSPIEBApreference programs extended to the LDCs, resist objections raised by the textile sector to the use of 'regional cumulation' allowed by the EU 78For further details and analysis, see: Baysan, Panagariya, and Pitigala (2004); Trade Policies in South Asia: An Overview, Chapter 5, (Volume 11), World Bank, 2005; Global Economic Prospects: Trade, Regionalism, and Development, World Bank, 2000. 79For details, see: Baysan, Panagariya,and Pitigala (2004). towards meeting the latter's 'rules of origin' conditions. This will help Bangladesh's RMG exports in the EU markets;and Lobby strongly in the USAfor duty-free entryfor its RMG exports, since such a preference has already been extended to the LDCs in Africa under the Africa Growth and OpportunityAct. Since many export items do not benefit from the duty-free access to imported inputs through the bonded warehouse and EPZ schemes, it is critical to ensure a wellfunctioning duty drawback system such that rebates are paid quickly. To this end, the full implementation of the new duty drawbackand bonded warehouse scheme is needed immediately. 4.29 Recommendationsfor the regional track to trade liberalization throughfree trade area (FTA) First, continue with unilateral trade liberalization to minimize the likely adverse trade diversion effects, the key elementsof which are listed above. Both for the SAFTAand for any bilateral FTA that might be signed: o minimize sectoralandlorproduct exception from such FTA agreements; o have clearrules againsttariff-rate quotas; o have 'rules of origin' that are liberal, simple,transparent,and uniform for all products; o extend suchagreementsto servicestrade and investment; It would be desirable if countries in the region use the instrumentality of SAARC to promote economic cooperation in areas other than trade, de-linking those with cooperation on trade under SAFTA. Using the SAFTA instrumentality to promote trade and other agendas could face the risk of each being held hostage to the progress in the other. Specific areas of regional economic cooperation where the SAARC forum could be used include: infrastructure;trade facilitation; and harmonization of technical and sanitary and phytosanitary (SPS) standards in line with the standardsof the major export markets. 80For an analysisof economic arguments onthe topic, see Baysan,Panagariya,and Pitigala (2004). 56 Chapter 5: Urban Development and Economic Growth in Bangladesh I. Urbanizationand Urban Concentrationin Bangladesh: A Process of Rapid Transformation 5.1 Bangladesh, a predominantly rural country, is undergoing a transformation toward urbanization at a remarkable pace. Its urban population has grown at a yearly average rate of 6 percent since independence,at a time when the national population growth was 2.2 percent. As a result, urban population has grown six-fold, compared with a 70 percent increase in rural population. As per recent UN data, about 35 million people, or approximately25 percent of Bangladesh's population, currently live in urban areas, compared to only 8 percent at the time of independence: the number is projected to cross 80 million by the year 2030 (Figure 5.1)." The share of urban activities in GDP increased from 26 percent in 1972-73 to over 42 percent by 1998-99,82a period over which agriculture sector's share in GDP fell from 60 percent to 26 percent. Figure 5.1: Urbanizationin Bangladesh Uban Population(%) 180 160 n -g140 = 120 E. -g - loo 4 80 oP 60 n 40 20 0 Data Source: UN World UrbanizationProspects 5.2 Urbanization has been more concentrated in the largest cities. More than half the urban population lives in the four largest cities (see Table 5.1). Dhaka, the national capital, is the largest of these with 12 million resident^:'^ an eightfold increase in its population since 1970. Its population is projected to reach a staggering 22 million by the year 2015 according to UN estimate^.'^ This would make Dhaka, currently the world's 1lh largest city, the 6h largest city in the world and 31dlargest in Asia by 2015. The next three largest cities - Chittagong, Khulna, and Rajshahi - are also growing rapidly, each having experienced at least a five-fold increase in population since 1970 (ten-fold increase the case of Rajshahi). At the other end of the size spectrumare about 300 smaller urban areas, which account for just 4 percent of the urban population.85 Economic activity is just as concentrated in the largest metropolitanareas. Urban growth in Bangladesh is considerably higher than its South Asian neighbors - urban populations in both India and Pakistan have grown at about 3 percent annually since the 1970. Similarly, urban population growth in Brazil has increased by 3.7 percent a year since 1970. 82Center for Policy Dialogue (CPD) (2001). 83This is about one-third of the national urban population. 84Ahmed et al., mimeo, World Bank, 2006. Chowdhury (2004). 5.3 Urbanization and concentration of it in the largest cities is a byproduct of economic development, aphenomenon hardly unique to Bangladesh. Developed countries, almost by definition, have already made the transition to having a vast majority of their population living in cities. Similarly, as the emerging economies, including Bangladesh, have moved up the development ladder, the share of urban activities in GDP has increased and a rapid shift toward urbanization has resulted. Fast growing economies such as Korea and Malaysia, in fact, have doubled their urbanization rates to 65 and 80 percent, respectively, since 1970. 5.4 The urbanizationprocess has important implicationsfor productivity and growthprospects of a country, and the associated opportunities and challenges require careful management. Production typically benefits from being located in densely populated urban areas that provide a pool of skilled workers, a network of complementary firms that provide backward and forward linkages for supply chains, opportunities for knowledge flows, and a critical mass of consumer^.^^ Presence of economies of scale and scope that arise in the early development stages generates increasing returns, with benefits for growth and productivity. At the same time, concentration of urbanization in a few large cities without concomitant improvements in their governance and infi-astructurealso carries associated costs in the form of congestion, pollution, and transport and services bottlenecks. After a point, the costs tend to dominate and decreasing returns set in, resulting in lower GDP growth, as shown by recent cross-country studies.87 Accordingly, policies to manage this challenge need to first ensure strong governance of the larger cities to benefit from their dynamism and agglomeration economies, while also creating a level playlng field for the smaller cities, to enable their emergence as viable urban alternatives. The latter focus is important to reduce concentration pressures on the larger cities and spread the agglomeration benefits more widely. 5.5 In this context, two aspects of the urbanization process in Bangladesh seem quite relevant. First, from aproduction perspective, how can the net benejitsfrom urban concentration be maximized? Second, from a consumptionperspective, how can cities overcome administrative andjiscal constraints toprovide services that are valued by households andjirms? The two issues are necessarily intertwined. The first involves national level policies such as expenditure on inter-regional transport network and decentralization that shape the patterns of urban concentration and have a bearing on demand for urban services. The delivery of city services, itself along with improvement of other aspects of urban city management (such as revenue mobilization), works to reduce the costs of urban concentration and hence improve the net benefits of urban concentration. Additionally, improvement of urban services in the smaller cities has the spillover effect of alleviatingpressures on the largest cities. To address these issues, it seems important to examine more closely the costs and benefits of urban concentration in Bangladesh, including the factors driving it, and also the structure of urban management (including division of responsibilities at the national and local levels) that guides the extent to which individual cities can innovate. The rest of this chapter is organized along such lines. 11. Urban Concentrationin Bangladesh: Extent, Causes,Benefits,and Costs Urban Concentration in Bangladesh 5.6 As already noted above, theprocess of urbanization in Bangladesh has been rapid and uneven: it has been much more concentrated in the largest cities, especially Dhaka (Table 5.1). The four largest cities - Dhaka, Chittagong, Khulna, and Rajshahi - account for over 60 percent of the urban population, up from 48 percent in 1970. One-third of the urban population resides in Dhaka alone. Dhaka, Chittagong, and Rajshahi have each outpaced national urban growth. The urban concentration trends, not surprisingly, largely mirror the concentration of economic opportunities in these cities. Most major 86 World DevelopmentReport, World Bank, 1999-2000. 87 For example, Henderson (2003). industrial activities and auxiliary businesses services are concentrated in the largest cities. Dhaka alone accounts for 80 percent of the garments industry-the mainstay of manufacturing in ~ a n ~ l a d e sInh . ~ ~ Dhaka and Chittagong, about one-tenth of the labor force is engaged in the garments industry, which is over seven times the national average (Annex Table 5.1). If this is linked to availability of complementary businesses services, then the data bear that out as well. The representation of business services, particularly finance and real estate in employment is considerably higher in the four largest cities relative to the rest of the country. At the same time, and not surprisingly, the share of population in the largest cities engaged in primary activities is significantly lower than in the rest of the country. Table 5.1: Population growth of the largest agglomerations City Population Population Population Population Population Annual Growth Rank (2000) (2000) (1990) (1980) (1970) (1970-2000) Dhaka 1 12300 6619 3248 1474 7.1% Chittagong 2 3581 2265 1333 693 5.5% Khulna 3 1426 972 622 310 5.1% Rajshahi 4 1016 517 238 105 7.6% Myrnensingh 5 328 189 108 N.A. Comilla 6 307 135 126 86 4.2% Note 1:Population in thousands; Note 2: Data are for agglomerations; Data Source: Global Cities Database http://www.econ.brown.edu~facul~henderson/worldcities.html 5.7 The degree of urban concentration in Bangladesh appears to be higher than in many of its comparators. A straightforward measure of urban concentration is primacy, or the share of the largest city in the total urban population in the country. In Bangladesh, Dhaka's primacy rate is about 32 percent, which is significantly higher than in neighboring South Asia countries (See Table 5.2) and is also high relative to countries with similar land areas - for example, primacy is 21 percent for Bulgaria and 23 percent in Korea. However, Dhaka's primacy seems in line with countries such as Honduras when these were at similar levels of economic development. Table 5.2: Urban concentration in comparator counties (%) Data Source: WDI Tables,SIMA Causes of Urban Concentration:Market Access, AgglomerationEconomies, and Policy Biases 5.8 Analyses for other large developing countries have identified the following factors as being key to firms' location choices across sub-national r e g i o d 9 (i) the quality and cost of complementary utility services, including electricity, water and telecommunication; (ii) market access, or the geographical reach, given existing transport infrastructure; (iii) agglomeration economies as measured by the presence of firms in own industry and of firms in related industries; (iv) labor and other regulations; and (v) access to serviced land. Due to data limitations, for the case of Bangladesh we restrict ourselves to assessing the Dhaka UrbanPoverty: Land and Housing Issues. Draft paper, World Bank, 2005. 89Lall, Venables, and Redding (2005). role of market access and agglomeration economies, in addition to that of major policy biases toward the largest cities. TheRole of Market Access 5.9 Regional variation in market access, the distance from and the size and density of market centers in the vicinity of thefirm, is an important factor leading to urban concentrati~n.~ Firms are attracted to areas with better market access, which leads to concentration of production in these areas." In the case of Brazil, India, and Indonesia studies have found market access to be the most important factor in determining where firms locate.'* Market access and urban concentration appear to be closely linked also in Bangladesh. Figure 5.2 shows market access - which is a combined function of road network and its quality, travel speeds, and population density - to be the highest around the Dhaka metropolitan area. As one moves out of the conidors that link Dhaka to other major urban centers, market access drops off quite rapidly. This does not, however, establish in any way that market access causes urban concentration in Bangladesh. It may well be that urban concentration, arising from other factors, is leading to higher market access. A more reasonable reading would be that market access and urban concentrationreinforce each other in Bangladesh. Figure 5.2: Market Accessibility across Urban Centers Source: Deichmann 2003 TheRole of Agglomeration Economies 5.10 Agglomeration economies are another nrajor reason why industrial and commercial activities tend to cluster. The benefits of agglomeration economies take two forms. Firms benefit from co-locating with other firms in the same industry: a phenomenon also known as localization economies. Firms also benefit from presence of complementary activities or urbanization economie~.'~Firms that produce The classic gravity model is commonly used in the analysis of trade between regions and countries (Evennet and Keller (2002)). It states that the interaction between two places is proportional to the size of the two places as measured by population, employment or some other index of social or economic activity, and inversely proportional to some measure of separation such as distance. Davis and Weinstein (2003) show that good market access has a magnified effect on location decisions because fvms will locate in the larger market and export (bearing transport costs) to the smaller ones, rather than vice versa. 92Deichmann et. al, 2005; DaMata et. al. 2005; La11and Mengistae (2005). 93World Development Report, World Bank, 1999-2000. intermediate goods will tend to locate close to producers of the final good (a forward linkage). Downstreamfirms, similarly,will gain from locating close to their suppliers (a backward linkage). These linkages can also serve as a channel for informationtransfers. Firms that are linked through stable buyer- supplier chains often exchange ideas on how to improve the quality of their products or on how to save production costs. 5.11 Do agglomeration economies explain urban concentration in Bangladesh? It is highly likely that the concentration of garment activities in Dhaka and Chittagong generate localization economies, a draw for newcomers in the sector. To test whether the largest cities in Bangladesh also provide more opportunities for urbanization economies, we calculate a population-weighted specialization index, with the expectation that lower the value of the index for a given city, the less specialized is the city and greater is the presence of urbanization economies in that location. The index is calculated at the district level,using data from the 2003 Labor Force Survey. 5.12 We first group the districts into three size categories. The first category consists of the Dhaka metropolitan area with a population of more than 5 million. The second consists of the cities of Chittagong,Rajshahi and Khulna metropolitan areas, which have populations more than 500,000 but less than 5 million. The third category consistsof all the other district^.'^ The index is measured as follows: where Ej is the share of industry j in national employment, SF is the share of industry j in total employment of agglomeration i, and the sum is over k industries locally. The index measures for each industry how much the local production share differs from the national share. If all industries mimic the national share the index has a value zero and the city is perfectly diverse. A highly specialized city would have an index approachingtwo. 5.13 Estimates of the specialization index suggest a more significant presence of urbanization economies in the largest cities. As seen in Table 5.3, relative to the smaller cities (size less than 500,000), the index has a lower value for the four largest cities in Bangladesh, suggesting diversified economic structures in these cities. This, together with the concentrationof the major economic activities such as garments production and organized business services in these cities, reflects scope for both aspects of agglomeration- localizationeconomies and urbanization economies- in the largest cities. A somewhat surprising finding,however, is that Dhaka is more specialized than the next three citieson both specialization measures.g5 On the other hand, the small and medium size cities appear to be more specialized, consistent with empirical evidence for other countries which similarly find smaller urban centersto be specializing in production activities such as food and beverages, textiles, shoes, or pulp and paper products, among others.96Therefore,while there is scope for localizationeconomies for the smaller cities in Bangladesh, the scope for urbanization economies is much more limited compared with the larger metropolitans. This appears to be another likely reason behind the uneven pace of different-sized citiesin Bangladesh. 94 Rural parts of the district have been excluded. 95 Much of Dhaka's excessive specialization comes from the fact that it diverges from the national share of agriculture employment (11 percent vs. 60 percent nationally) giving it a high overall index, and its share of garments employment is twice the national share, thereby giving it a high value in the manufacturing index. % Henderson et. a12001. Table 53: SpecializationIndex City Size 1 Manufacturing I Overall 5 million + (Dhaka) I 0.19 I 0.28 500,000- 5 million (Chittagong,Khulna, Rajshahi) 0.07 0.2 less than 500.000 0.4 I 0.5 L I I I Data Source: Labor Force Survey, 2003 TheRole of Policy Biases 5.14 Policy bias and resultingfavoritism also aid urban concentration. In a centralized governance structure, as in Bangladesh, the national government tends to favor one or two cities, typically the national capital (Bangkok, Dhaka, Mexico City, Jakarta, or Seoul) or certain financiaVcommercia1 capitals (Sao Paulo), over others. Such favoritism often involves disproportionate allocation of public expenditures toward the favored cities. These cities also receive disproportionatelyhigh levels of public services, while hinterland cities are restricted in their ability and autonomy to deliver public service.97 Favoritism can also involve the national government choosing not to invest sufficiently in interregional transport and telecommunications, so hinterland cities are less competitive location^.^^ Both forms of policy bias are visible in Bangladesh. Policy bias can also take the form of restrictions in capital markets, exporttimport markets, and licensing of production rights, all favoring firms that locate in the national capital, although there is no direct evidence of this in Bangladesh. An additional problem in Bangladesh is the inefficient functioning of the national land developmentmarkets which, together with lack of fiscal decentralization, prevents land developers and local governments from actively developing alternate urban locationsand spreading developmentacross the urban hierarchy. Negative ExternalitiesArising from Excessive Urban Concentration 5.15 Whilemarket access, agglomeration economies, andpolicy biases influence location decisions of firms towards large cities, excessive urban concentration in the absence of concomitant improvements in urban management and infrastructure carries associated economic costs. Such costs are becoming fairly evident in the major urban areas of Bangladesh, especially Dhaka. Real estate in Dhaka has gotten out of reach for most, rapidly worsening traffic congestions have increased commute times substantially,Dhaka's infrastructureis being stretchedto the limit, and citizen ratings of most urban services border on complete dissatisfaction. 5.16 The cost of real estate in Dhaka has reached astonishing levels. Residential land values in prime locations of Dhaka are higher compared to thosefound in developed countries.99 For example, the Bangladesh Institute of Planners estimates that land in Dhanmondi, among the more expensive areas in Dhaka, is Tk 3,5001ft2 or roughly US$50. Similarly, land in Gulshan is priced at Tk 3,194 lft2 and Baridhara at Tk 3,000 lft2. Even in peripheral areas of Dhaka, such as Uttara, Pallabi, and Shyamali,land costs over Tk 1,500Ift2 (over US$20). Consider in comparison the land prices (in) in the cities of the United States (whose per capita income in purchasing power prices is about 20 times higher): US$13/ft2 97However, it is quite likely that there are increasing returns in the provision of many public goods. For instance, if a country is to build 100 miles of paved roads, it may be efficient to build most of it joined together in a dense network rather than scattered around. In this case an, 'favoring' large cities in public expenditure and provision may be an eflicient allocation of resources. 98 Recent literature contim the importance of inter-regional discrepancies in influencing urban concentration. Henderson (2003) finds that investment in national roads and highway systems significantly reduces national urban primacy. Gallup, Sacks, and Mellinger (1999) and Rosen and Resnick (1980) find similar results for investments in waterways and railways. 99Seraj and Afiin (2003) in Boston; US$14/ft2 in Chicago; US$30/ft2 in Los Angeles; US$ll/ft2 in Miami; US$32/ft2 in New York; US$26/ft2 in San Diego; and US$64/ft2 in San ~rancisco.'~~The situation is not that much different in other metropolitans in Bangladesh. Land prices in Khulna have been estimated to be similar to those in Dhaka, and are about 15 percent lower in chittagong."' The least expensive middle class housing in the country is estimated at about Tk 700,000, which is more than 10times the median income. 5.17 Dhaka's high land prices stem from the topographicalfeatures that limit its supply of land as well asfrom poor land management. Dhaka is located in a flood plain and surrounded by rivers which periodically flood the lowlands around the city. This creates scarcity of developable land, reflected in the city's high population density of about 20,000 persons per square lulometer, going up to 100,000persons in some parts.'02 The land constraint imposed by the natural features is exacerbated by poor land management by the Rajdhani Unnayan Kartripakkha (RAJUK), the regional planning and development agency. Since its formation in 1959, RkTlTK has developed serviced parcels sufficient only for 16,000 housing units, or less than 400 units per year. At the same time, the area within its jurisdiction has added more than 6 million people-roughly 1 million household^.'^^ Thus, the one public agency responsible for housing infrastructure,and with the means to assembleparcels, has been able to meet one 1-2percent of demandfor serviced land. 5.18 Infrastructure and service bottlenecks in the major urban areas have becomejust as acute. In a 2002 consumer satisfactionsurvey (Table 5.4),lo4less than 20 percent of households in each of the four metropolitan areas expressed satisfaction with almost any urban major service. Police, land registration, public transport, electricity, and especially judiciary were rated very poorly. In Chittagong, only 5 percent of the respondents were satisfied with the quality of education and 4 percent with the quality of health care. Table 5.4: Satisfactionwith Services(% of surveyed households) Services Dhaka Chittagong Khulna Rajshahi Police 2 0 1 2 Land Registration 2 1 10 4 [ ElectricityServices I 8 1 2 1 12 1 2 I Judiciary 8 1 2 5 Health Care 11 4 18 9 Garbage Disposal 15 10 12 10 SeweraeeISanitation " 17 16 11 16 Education 1 21 1 5 1 28 1 12 1 I Drinking Water 1 27 1 9 I 11 I 8 I Source: Proshikha (2002) looGlaeser and Gyorko (2003) lo'Housing Finance Rdonns in Bangladesh, SouthAsia Finance and Private Sector, World Bank, 2004. lo2Islam (2004). '03Housing Finance Rdorrns in Bangladesh, SouthAsia Finance and Private Sector, World Bank, 2004. '04The survey was conducted in 2002 by Proshikha (a BangladeshiNGO) and the Public Affairs Center (Bangalore, India). Six satisfaction-relatedresponse options were provided to the households ranging from stronglysatisfiedto stronglydissatisfied. These were aggregated into three groups: (i) "satisfaction", which representedthe share of respondents strongly satisfied and moderately satisfiedwith any particular service; (ii) "dissatisfaction",which representedthe share of respondents stronglydissatisfiedand moderatelydissatisfied with any particular service; and (iii) "ambivalent", which represented the share of respondentsmarginally satisfied and marginallydissatisfied. 5.19 Firm surveys, too, reflect deep dissatisfaction on the part of business owners and executives with infrastructure and access to land in the main cities.lo5 Firms in Dhaka and Chittagong were equally unhappy with the quality of electricity supply, reporting power interruptions of about 250 days per year. Forty percent of firms interviewed in Dhaka reported access to land as a severe constraint to business expansion. 5.20 Until recently, Dhaka ranked among the world's most polluted cities, with emissions from vehicles with two-stroke engines (baby taxis, tempos, motor-cycles) and trucks and buses being the main contributors to air pollution. However, with the phasing out of two-stroke engines on December 31, 2002, air quality has improved quite dramatically. PM2.5 concentration in Dhaka declined from 250 micrograms/m3 prior to the phase-out to 150 micrograms/m3 on Jan 2, 2003.1°6 However, the overall welfare effects of this policy is not clear as the phase out baby taxis are likely to be used in other cities and increaseemission levels in those places. GrowthImplications of 'Excessive' Urban Concentration 5.21 The relationship between urban concentration andgrowth appears to have an inverse-U shape, for given levels of income and urban governance and infrastructure. As noted, concentration of resources in early stages of development is useful and improves efficiency as economies of scale and scope help get the most out of scarce resources. At the same time, the economic costs - congestion, overburdening of infrastmcture etc. - also begin to rise with increasing urban concentration, more so in absence of good urban management. In fact, recent research suggests that there is an optimal range of urban concentration relative to the quality of the country's urban governance and infrastr~cture.'~'As show by Henderson (2003), for example, deviations from this optimal point lead to productivity and growth losses. 5.22 Primacy of Dhaka is considerably higher than the "optimal"primacy level suggested by Henderson's cross-country estimations. At Bangladesh's income level and urbanization scale, Henderson's estimates suggest an 'optimal' primacy rate of around 21 percent for Dhaka, whereas it is about 32 percent, or more than two standard deviations higher than the optimal. This implies a loss of at least two percentage points in annual GDP growth. This likely overstates the case, but does make the point that excessive concentration of economic activity in one city, Dhaka, relative to its level of governanceand infrastmcture, could have serious growth costs for the nation as a whole. III. Urban Management Structure and Financing and Delivery of Local Amenities 5.23 In the previous section, we considered the benefits from agglomeration economies and the costs associated with excessive urban concentration. To be sure, the extent of the negative externalities is related to the quality of the city management. If cities suffer from bad management and inadequate resources, then that will only add to the bottlenecks in infrastructure and service provision, resulting in diminishingreturns from urbanization setting in sooner. On the flip side, better management of cities can help contain the costs of increasing concentration, prolonging the period of net benefits from urbanization. The quality of urban management in smaller cities also matters -the higher is the quality, the lower will be the concentrationpressures on the major cities. What is the structure and state of urban management in Bangladesh? '05 The Bangladesh Enterprise Institute and the World Bank carried out the investment climate survey, covering 1001 firms in Dhaka and Chittagong. '06Details on air quality issues are at http://www.cleanairnet.org/caiasia lo'For example Wheaton and Shishido(1981), and Henderson(2003). 64 5.24 Out of the 522 urban areas identified by the 1991 Census Commission,only 295 urban centers have urban local governments. The six largest cities have a city corporation status, while 289 other urban centers have Pourashava or Municipality status. Pourashavas are further classified into four categories based on their own source revenue potential. These are Special Category Pourashavas (only 2 in number), Category I Pourashavas (61), Category I1 Pourashavas (46), and Category I11 Pourashavas (190). In centers with no local government, urban infrastructure and services are provided by the field administration of the central government through its line ministries.Io8 5.25 City governments in Bangladesh do not have much autonomy to develop economic development programs or design and provide most public services. There is considerable vertical imbalance where most 'urban' functions are the responsibility of the central government. This includes all utility senices, housing and land development, secondary and higher education, and economic development planning. Several national ministries are responsible for the management of the 'urban sector'. The Ministry of Local Government, Rural Development and Cooperatives (MLGRDC) and the Ministry of Housing and Works (MOHW) share the responsibilities of developing and providing urban senices and infrastructure through the Local Government Engneering Department (LGED) and the Department of Public Health Engineering (DPHE) respectively. Urban development authorities such as RAJUK are part of the MOHW. Local autonomy is further stifled by the fact that local governments have little or no choice on the staffing, nor do they have control over the wages for their employees. Further, key personnel at the local levels are central government employees with limited accountability to residents. 5.26 City corporations and Pourashavas are responsible for delivering a set of generic public services, which leaves little scopefor competition among them to attract investment and labor. Table 5.5 summarizes various responsibilities of local governments and other service providers along with their main revenue s~urces."~It also seems that by creating many small Pourashavas, the government has not carefully considered scale economies in service provision or the implications of limited fiscal and administrative capacity at the local levels, creating dependence on fiscal transfers from the center."' 5.27 City governments are unable to ensure satisfactory delivery even of the small subset of responsibilities entrusted them, due to their weakfiscal and administrative capacity. Importantly, they do not have buoyant tax handles, which limits their own-source revenues. The main own-source revenue for Pourashavas is the property tax and associated fees for water, lighting and conservancy that are imposed on the same base, collectively called the holdings tax. The base of the property tax is the 'annual value' (or rent) of buildings and lands, as is common in many other countries."' Rebates on the '08Chowdhury(2004). Io9 There are only small differences between the responsibilities of a City corporation and a Pourashava. However, in case of Dhaka and Chittagong city corporations, water supply and sewerage, planning and development responsibilities have been given to special development authorities (RAJUK for Dhaka, CDA for Chittagong, and WASA for both Dhaka and Chittagong). Water supply is a local governmentresponsibilityfor the Pourashavas. "O Asaduzzamar (1996), Chowdhury (2004), and Bangladesh Municipal Finance Management Sector Study, InfrastructureOperationsDivision, SouthAsian Region World Bank, 1998. "' The tax on land and buildings is imposed at a maximum rate of 7% of the tax base (the annual value); and governmentregulationsalso set the maximum rates for the other parts of the holdings tax, at 3% for the lighting rate, 7% for the conservancyrate, and 10% for the water rate, giving a maximum combined tax rate of 27%. This overall rate of taxation is similar to, if slightly higher than, other countries with an annual value system of property taxation Bangladesh Municipal Finance Management Sector Study, World Bank, 1998. annual taxable value are offered to owner occupied properties to deduct mortgage costs as well as for maintenance e~~enditures."~ Table 5.5: Providers responsiblefor deliveringurban services Name of Agencies / Main Services I Sources of Finance City Corporations 1 Sanitation,solid waste disposal,road I Property taxes, conservancy, building and maintenance]street lighting, lighting and water rates, fees, traffic signaling,parks, playground, fines, rental income, poverty alleviation,slum improvement government grants, donor fiinds. Pourashava Sanitation,solid waste disposal, Road Property taxes, conservancy, building and maintenance,Street lighting, lighting and water rates, fees, traffic signaling,parks, playground, fines, rental income, poverty alleviation,slum improvement, I government grants and donor Planning, etc I funds. Water and SewerageAuthority Drinking water supply and sewerage 1 Water tariffs, loans, grants (WASA) in Dhaka and Chittagong I from govt. and donors Development Authorities: Planning and development of physical I From sale of lands and RAJUK, CDA, KDA, RDA infrastructure 1 government grants Departmentof Environment EnvironmentControl Governmentbudget Specializedauthorities: Civil works, Housing,physical Governmentbudget and PWD, NHA, DPHE & LGED Development donor's grant and aid Source: Chowdhury (2004) based on dzflerent sources 5.28 Weak administrative capacity and strong political interests limit the extent to which local governments can tap an expanding tax base and enforce compliance with taxes. In principle, property tax collection should be sufficient to finance delivery of services expected of local government. This, however, is not the case. While the tax base is scheduled for revaluation every five years, the reassessments are seldom performed in practice. The administration of the stagnant tax bases is also fraught with problems. The rate of tax collection across Pourashavas is low -between 2000 and 2002, only 29 percent of smaller Pourashavas collected more than 50 percent of their estimated holding tax. Further, only 25-30 percent collected more than 50 percent in arrears.Il3 Moreover, in many local governments, the assessed base and taxable values are recorded by hand, which increases the potential for rent seelung and limits the potential for regular updates.Il4 Growth in revenues from the property tax is further limited when local politicians make promises of not raising tax rates (as was done in the 1990s). Finally, the linkage between taxes and the receipt of public services is extremely weak, which lowers tax payer confidence and willingness to pay property taxes. 5.29 In the absence of strong own source revenues, most local governments (especially Pourashavas) rely on central government transfers. These transfers often account for 50 percent of local government revenues. However, such grants, in turn, depend on the central government's own fiscal wellbeing; thus their reliability as source of municipal revenues is limited. There has also been a steady decline in grant allocation per Pourashava over time. With additional Pourashavas being created on a regular basis, total government grants per Pourashava fell by more than half between 1996and 2002 (Table 5.6). This has adversely affected the ability of municipalities, especially the smaller ones, to provide servicesto their citizens. ' I 2 Maintenance rebates are offered regardless of whether the resident or the landlord has incurred these expenditures. ' I 3Similarproblems of low collection rates and incompleteassessmentsof the property tax are seen in Indian cities. 'I4 Bangladesh Municipal Finance Management Sector Study, World Bank, 1998. Table 5.6: Reductions in central government transfers Source: Ministry of Local Government, GOB,Dhaka; Chowdhury(2004) 5.30 The transfers from the central government to urban local governmentsare mainly in the form of: (a) specific transfers for investment expenditures identified in the Annual Development Plan (ADP); (b) block grants to finance local expenditures; and (c) revenue budget transfers to finance recurrent expenditures, mainly salary subventions for municipal employees (City Corporations are not given salary subvention grants). Grants to local governments in support of the ADP are declining. In the year 199711998, City corporations and Pourashavas received 0.82 and 0.86 percent of total ADP grants respectively, and these figures fell to 0.63 and 0.74 percent by 200112002. Two important issues in the context of ADP allocations are not clear. First, the mechanism and criteria by which ADP block grants are allocated across different types of local governments are not defined. Second, the rationale for centralizedadministrationof the bulk of ADP block grants is unclear. Further, even when ADP grants are locally administered, they are pre-committed to specific sectors, which constrains innovation at the local level. IV. Policy Options To Improve the Contribution of Cities to Economic Growth: Improving Management of Cities and RemovingPolicy Distortions 5.31 The urban management challenge is twofold First, to improve access to and quality of basic services and infrastructure in Dhaka and other major metropolitans. Second, to improve theprospects for the smaller cities to emerge as viable urban alternatives. Cutting across both challenges is the need to devolve key services to city governments, together with clear delineation of duties and accountability chains between the federal and city governments, among the various agencies involved with urban management functions, and between the service providers and the citizens. This has to be carefully sequenced with improvements in city governments' ability to perform, particularly by enhancing their own revenue sources and building their technical and administrativecapacity. To improve own revenues, local governments first need to step up collection from existing tax bases, for which training of the appraisers and computerizationof billing would be useful. However, these efforts can succeed only if the link between local taxes and quality of public services is strengthened. 5.32 A multi-pronged holistic approach is necessary to effectively tackle urban management problems of the severity and scope seen in ~haka."' Strengthening the city's governance structure is theforemost priority, and would involve addressing the mentioned issues of decentralization, capacity and resource constraints, and fragmentation of responsibilities and accountabilities among the various agencies. Some specific recommendationsfollow: 'ISAhrned et al., rnimeo, World Bank, 2006. Strengthen the role of the Dhaka City Corporation (DCC) in urban management and improve inter agency coordination between the DCC and various development authorities in charge of infrastructure development and service provision. The DCC does not manage the use of public land nor does it have power over public utility companiesthat are operating within itsjurisdiction. There are more than 16government~autonomousorganizations directly involved with urban development in Dhaka, with another 30 other organizations having indirect roles.116Many of these are line ministries of the central government who have very little accountabilityto the elected local government. Improve DCC's own-source revenues. Dhaka's property tax base has not been reassessed for over 10 years. Furthemore, data from the Bangladesh Bureau of Statistics show that in 2001 the DCC collectedholding tax from only 180,000households out of the 1.1 million holdings in itsjurisdiction. It is critical to address fundamental issues of updating property tax rolls, computerizing billing and collection systems,and strengtheningenforcement. DCC has recently tried to initiate self-assessment of the holding tax to reduce administrative costs, which should bepromoted. The self-assessment system not only reduces the costs of administration and reporting, it also reduces opportunities for infomal agreements arrangements between homeowners and appraisers. According to municipal finance experts and local government officials, self-assessment system could increase property tax revenues by 40 percent.117 However, implementationof the self-assessmentprogram has been stalled following a High Court stay order in response to petitioners who argued that the program did not have provisions of appeal in case of dispute over assessment. Civil society organizationshave also opposed the self-assessment system in absence of a guaranteed link to public service standards. 5.33 In addition to improving governance of Dhaka, it is also important to consider options for enhancing thepotential for other cities as viable investment destinations. As in the case of Dhaka, the emphasis should be on improving the governance structure of these cities - a multifaceted agenda involving developmentof their administrative capacity, strengtheningof their own-revenues, and greater autonomyover local matters suchas utility services, crime prevention etc. 5.34 A carefully thought-out national urban development strategy is needed for more balanced urban development across the country. This would seek to remove policy biases - such as those in public expenditure allocations and public investments in transport networks - that work against the emergence of smaller cities, and to create incentive structures and competitive pressures among cities such that (public and private) investment and labor movement are guided by objective measures of urban management. Investments in inter-regional transport and communicationsare likely to make the smaller cities more attractive to private economic activity and reduce excessive concentration in the largest cities."' To be really effective, concomitant improvements in local infrastructure and service provision "6 Center for Policy Dialogue (CPD) (2001). "'Chowdhury(2004). "'While infrastructurelinks may enhance productivity and, in combinationwith complementary investments, attract firms to smaller cities in the hinterland, these investments can also have unintended consequences. In particular, inter-regionaltransport networks provide firms in the larger cities easier access to lagging-region markets, in competition with local producers. Higher transport costs for these large-city producers are often more than made up for by their inherent advantages such as economiesof scale, better know-how and lower unit cost of production. This could result in a flow of resources and opportunities out of the lagging region. within the smallercitieswould also be important,llgwhich, in turn,will require better governance of these citiesand large public investmentsby the local governments.'20 5.35 The recently initiated Municipal Development Funds (MDF) appears to be a useful way to address the large investment needs of the local governments. The MDF, a government-owned company, was established in 1998with initial funding of US$70 million from the World Bank to provide financial support to Pourashavas for infrastructure development and service provision - including, water supply, roads, drainage, sanitation, solid waste disposal, street lighting, development of office complexes and community centers. The fund explicitly requires Pourashavas to improve performance on many counts before they become eligible to apply for its resources. In particular, holding tax collection rates need to cross 50 percent of estimated revenues; the municipality's accounting system should be converted to a double-entry system; office management and accounts should be computerized; and local stakeholders should be involved in choice of projects. Currently, around 100 municipalities have applied for MDF. An important implementation issue relates to development of the Pourashavas' capacity to enable them to quality for the MDF. Otherwise, the less developed Pourashavas with weaker capacity risk being left behind. 5.36 Finally, land and housing market reforms also need to be addressed to accommodate new and existing demand. RAJUK's limited capacity as a land servicing agency needs to be enhanced - its mandate needs to be updated and performance improved. At present physical constraints in land availability are worsened by widespread public ownership of land, poor land record maintenance, costly registration procedures, and unclear property rights. These issues will need to be addressed to improve the ability of citiesto accommodategrowth. 5.37 In sum, Bangladesh would need to unleash the enormous economic potential of its major urban centers to ensure sustained high growth rates. While special attention is warranted toward Dhaka, which would remain an important engine of growth, a more enabling environmentis needed other urban centers to emerge as viable alternativesto Dhaka. This will require long-termcollaborative efforts of central and local governments, development agencies, internationaldonors, the private sector, and local communities. The main strategicissuesthat can help start-offthe process include: Building on agglomeration economies in large metro areas by improving management of 'urban' functionsand promoting inter-agency coordinationamong serviceproviders; Reducing vertical imbalances (in responsibilitiesand revenue generation)and building local capacity. Improving local fiscal capacity and linking improvements to service delivery (the MDFs appear to be a useful idea). Improving attractiveness of alternate locationsthrough regional transport improvements. Reducing growth constraints from land supplyand infrastructurebottlenecks. 119Paul Krugman (1991). lZ0 Recent empirical evidence on Brazil (DaMata et. a1 (2005)) shows that along with national policies such as regional transport networks and education quality, local matters such as land use, zoning, and crime reduction, too, are importantfor a city's growth. Chapter 6: Sub-NationalIncome and Expenditure Dynamics in Bangladesh: Insights from the Macro-GDP and Micro-Survey Data I. Introducingthe Theme 6.1 Bangladesh is geographically a small country with ethnically much more homogenous population than other countries in South Asia. Nonetheless, historically it is marked with considerable regional differences in dialect, custom, agrarian relations and social development.121 Although the presence and persistence of differences in human development indicators in the country are often discussed,122regional differences in income and other economic indicators are less known. This chapter seeks to advance the understanding of the regional income and expendituredifferences. 6.2 Income differences at the sub-national level matter in large part because of their impact on income growth at the national level. For one, the growth potential of a nation would not be fulfilled if any of its major geographicaVadministrative units lags its own potential. Lagging regions also create clear demarcations between the "haves" and the "have-nots," which, over the longer run,can sow the seeds for social conflict andjeopardize sustainabilityof good growth at the national level. 6.3 The mainpolicy challenge, however, is to unleash the growth potential of each region, and not pursuit of regional equality as a goal of intrinsic worth in itself. Single-minded pursuit of regional equality can, in fact, be counterproductive - it is likely to retard the better performing areas without necessarily lifting up the lagging ones. Similar to nations, regions, too, grow toward their own economic potentials, which depend on a host of economic and social factors - some in the control of policy makers and some not. Also, as economies grow and diversify there is a tendency to reap the benefits of agglomeration economies, spatial externalities, and increasing returns to scale leading to the formation of large urbanlindustrial clusters and regional specialization, with implications for increasing spatial di~er~ences."~The growth potential of the poorer regions would very much depend on informed policy choices that encourage them to benefit from the "externality-generating" activities, and facilitate flow of resources (especially manpower) to the parts of the country that are economicallymore dynamic. 6.4 This chapter seeks to shed light on the sub-national differences in the levels and growth of income and expenditure in Bangladesh, although within the confines of fairly restrictive data limitations. The chapter also identifies some factors correlated with crossdistrict differences in per- capita expenditure growth over FY1992-2005. Specifically,expenditure growth is found to be negatively correlated with the initial levels of expenditure inequality, gender disparity, and large landholdings, and positively correlated with the degree of urbanization and access to electricity. Also, in general, districts with lower per-capita expenditure experienced faster expenditure growth, implying mild regional convergence in per-capita expenditure, although the speed of convergencewas quite slow; the estimated time to full convergenceis about 90 years. 6.5 These result are indicative at best, as the expenditure data are not representative at the district level. Nonetheless, per-capita expenditures and per-capita GDP are tightly correlated across districts for 12'Two Bengals had historical differences in geography and subsequent social formations (Eaton (1997); Khan (2002)). However, the significant differences were observed even within Eastern Bengal, especially in tenns of agrarian systems (Bose (1986)); between agrarian frontiers and settled territories (Ludden (2004)); and in respect of customs and traditions (Roy 1988). BIDS (2001); Sen and Ali (2004); World Bank 2004. 123Rosenstein-Rodan (1943); Krugman (1992); Hoff (1998); Jalan and Ravallion (1998); Kanbur and Venables (2005). the years for which comparable data are available. In light of that, there is some value in identifying the policy and institutional correlates - that are representative of the individual districts - of district level expenditure patterns. The exercise adds another dimension to better understanding the nature of the aggregate growth process in Bangladesh. Still, the analysispresented here is not developed enough to be able to be translated into a concretepolicy message. 11. Income, Growth, and Inequality Variations Across Bangladesh 6.6 Bangladesh is divided into six Divisions, each Division is sub-divided into Districts or Zilas: there are 64 such Districts in all. The average population of each District is about 2 million with an average land area of about 2,250 sq.krn. The Districts are fbrther sub-Divided into Thanas (for rural areas) and one or more Pourashavas or Municipalities: in all, there are 289 Pourashavas and 6 City Corporations (Chapter 5). Thanas are sub-divided into Unions and Unions into ~ i l l a ~ e s . ' ~ ~ 6.7 Two sets of data are used here in this chapter. One is the District-level macro GDP data from the 2000 Statistical Yearbook of the BBS. The other is the micro-level "consumption expenditure" (henceforth "expenditure") data derived from the Household Income and Expenditure Survey (HIES) and aggregated at the District level. Box 6.1 provides more details on the two data sets. GDP data are also used for the 20 Old Districts for longer inter-temporal comparisons. Box 6.1: RegionallyDisaggregatedData: Macro GDP Vs. Micro ExpenditureData Regional income or expenditure level or growth data are difficult to come by in Bangladesh -the state of affairs for social indicators or agriculture production data is better. The BBS published GDP data at the level of the 20 Old District level between FY1982-99, but discontinued that subsequently. The BBS recently also published GDP data for the 64 Districts during 1996-2000,but then discontinued that as well. The GDP data, therefore, do not allow for any detailed analysis of growth trends at the subnational level. The use of micro data on "consumption expenditure" aggregated at the District level then becomes a potentially important avenue for exploring growth dynamics. Their main advantage over the GDP data is that they cover a longerperiod, FY92 to FY05, deriving from the HIESs done in FY1992, FY2000, and FY2005. The micro-level expenditure data are preferred over rnicro-level income data from the same data source because of lower measurement error (see, Deaton 1997). One problem, however, is that the district level aggregates are not representative of the districts in a statistical sense. Even under the best of circumstances, when micro expenditure data are, in fact, representative at the regional level, it is difficult to reconcile them with the national accounts (see Deaton and Kozel 2005 for a survey of issues in the Indian context). As a result, the regional rankings of GDP and micro expenditure data seldom match closely. The reconciliation problem can be even more pronounced for growth rates (see, for example, Azzoni et a1 (2005) for evidence on Brazil). Not surprisingly then there is less than one-to-one correspondence in the rankings of Districts by GDP and micro expenditure data. Nevertheless, there is significant congruence. The Spearman rank correlation between the GDP and expenditure rankings for the 64 Districts is 0.43 for the year 1996and 0.38 for 2000. Similarly, the correlation coefficient between District-level GDP and expenditure is 0.80 for 1996and 0.71 for 2000. GDP and ExpenditureVariations at the Divisional Level 6.8 Per capita GDP growth in the 1980s and 1990s was strikingly similar across the 6 Divisions (Table 6.1) -Dhaka, Chittagong,Khulna, Barisal, Sylhet,and Rajshahi. As per aggregated Old Districts data, annual Divisional growth varied in a narrow 1.8-2.8 percent range across the Divisions over the l Z 4"Role of UNDP in Promoting Local Governance and Decentralization in Bangladesh Report of the Initial SPPD ScopingMission, 2002. FY83-99 period. Similarly, as per aggregations from District level GDP data, Divisional growth in the second half of the 1990s varied within an even narrower band of 2.7-3.1 per~ent.'~'Another noticeable feature is that the growth acceleration at the national level since 1990was experienced in varying degrees by each of the 6 Divisions. It was most pronounced in Sylhet, where per-capita growth accelerated by over 3 percentage points during the FY90-99 period compared with the FY83-89 period. Dhaka and Chittagong, with 2 and 1.3 percentage point acceleration, respectively, also saw a significant pick-up in their growth rates. On the other hand, growth in Barisal improved by a small amount. At the same time though, afair amount of dispersion is seen among the Divisions in their per-capita GDP levels. In FYOO, the per-capita GDP of the most well-off Division (Dhaka) was about 50 percent higher than that of the least well-off Division (Sylhet). Table 6.1: GDP Levels and Grow h by Division Aggregated From District Level Data Aggregated From "Old District" Data GDP per capita GDP Share in GDP per capita (current prices) Total (7%) growth GDP percapita growth Division FYOO FYOO FY9640 FYS3-89 FY90-99 FYS3-99 Dhaka 22303 37.7 2.9 1.3 3.3 2.4 Chittagong 18128 19.3 2.8 1.7 3.0 2.5 Khulna 17875 11.6 3.1 2.6 3.O 2.8 Barisal 15383 5.8 3.0 2.1 2.2 2.2 Sylhet 14886 5.2 2.7 0.0 3.1 1.8 Rajshahi 15174 20.4 3.4 National I 18511 100 2 3 Source: BBS, and Staff calculations GDP and Expenditure Variations at the District Level 6.9 GDP data for the next administrative level, the District, are available for a relatively short period of FY96-00, although additional insights can be obtained from District-level expenditure data from the HIES, since they cover a longer period, ~ ~ 9 2 - 0 5As . ' ~in~the case of the Divisions, per-capita GDP growth was relatively uniform across the Districts. Almost 90 percent of the Districts experienced average growth in the 3-5 percent range during FY96-00, none of the Districts experienced negative growth and only two experienced growth of over 5 percent (Figure 6.1). Theper-capita expenditure growth rates, on the other hand, were much more varied, with average annualgrowth varying between -4.8 and 6.2 percent over FY92-05. Almost one-fifth of the Districts recorded negative expenditure growth over this period, while another one-fifth recorded positive but less than 1percent growth (Figure 6.1). 12'It must be noted that the data compiled by BBS at the level of the 20 "old districts" are of fairly weak quality. For one, these are based on the 1968 SNA, whereas the 1993 SNA is currently being used. The District level data, on the other hand, address some of these quality concerns. '26Although expenditure data are not representative at the District level, these are tightly correlated with GDP data across districts. Still, the District level aggregations are to be interpreted with caution. Figure 6.1: District-Level Variations in Per Capita GDP (FY96-00) and Expenditure GrowthRates (FY92-05) Source: BBS, HIES and Staff Calculations 6.10 Considerable dispersion in both GDP and expenditure levels can be seen among the Districts. In FYOO, the average per-capita GDP of the top five Districts was about two-and-a-half times that of the bottom five Districts. Per-capita GDP of Dhaka, the most well-off District, was three-and-a-half times that of the least well-off District (Gaibandha) in FYOO. Moreover, as seen in Figure 6.2, in FYOO, per- capita GDP in three-fifths of the Districts was less than Tk.16,000, while only four Districts - Dhaka, Gazipur, Narayanganj, and Chittagong-had per-capita GDP of over Tk. 25,000. The ratio of the top and bottom five Districts in terms of per-capita expenditure was also close to 2.5 in FYOO and in FY05. In FY05, per-capita expenditure in only three Districts was over Tk. 20,000, while in three-fifihs of the Districts it was less than Tk. 14,000. Annex Table 6.1 shows GDP and expenditure level and growth information for all Districts for which such data are available. Figure 6.2: District-Level Variationsin Per Capita GDP (FYOO) and Expenditure Levels (FY05) Source: BBS, HIES and Staff Calculations 6.11 Growth patterns across Districts show mild evidence of convergence in expenditure levels. In general, as also discussed later,Districts with lower initial expenditure levels in FY92 saw higher growth in their per-capita expenditure over the FY92-05 period. Comparisons between the richest and poorest Districts were even more striking. Average per-capita expenditure growth in the 5 Districts with the lowest per-capita expenditure in FY92 - Patuakhali, Gopalganj, Gaibandha, Netrokona, and Tangail - was 3.9 percent, compared with an average growth of 0.8 percent in the 5 Districts with the highest per- capita expenditure-Dhaka, Nilphamari, Sylhet, Chttagong, and Narayanganj. Furthermore, while six of the top 10 Districts in FY92 (in terms of per-capita expenditure) experienced negative growth over the FY92-05 period, each of the bottom 10Districts experiencedpositive growth. GDP Growth Variations at the Old District Level 6.12 The growth acceleration at the national level in the 1990s appears to have been shared across much of the country (Figure 6.3). Per capita GDP growth was higher in the FY90-99 period compared with the FY83-89 period in 14 out of the 20 Old Districts. All 6 Old Districts which experienced a growth slowdown in the latter period were among the seven fastest growing Old Districts in the FY83-89 period. Furthermore, the three slowest growing Old Districts in the FY83-89 period experienced the fastest accelerationin their per-capita growth in the latterperiod. Figure 6.3: Per-Ca~itaGDP Growth Across the Old Districts Source: HIES and Staff Calculations How "High"is RegionalInequality? 6.13 To estimate spatial inequality, we consider the contribution of variation in per capita expenditure across Districts to per capita expenditure variation across all individual^.'^^ This assesses to what extent regional inequality in Bangladesh accounts for the total inequality (i.e. inequality in the inter-personal income) in the country. The matter assumes special significance for the 1990s when the standard measures of inequality such as Gini index show sharp rise both for expenditure and income data.Iz8 6.14 The Theil index of inequality is well-suitedfor the analysis and is calculated in Table 6.2.129,'30 Three results are noteworthy. First, as with the conventional Gini index, the Theil measure of inequality shows a sharp increasein inter-personal inequality in expenditureduring the 1990s. Second,both within- District and between-District components of interpersonal expenditure inequality rose during this period with slightlyhigher pace of increase recorded for the latter. Third, and more importantly,the contribution of between-District inequality to total interpersonal inequality was relatively modest - between District 127On this, see Kanbur and Venables (2005). 128See, World Bank 2003 and Osmani et a1 (2003) for trends based on expenditure data; Khan and Sen (2004) for trends based on income data. 12'see, Bourguignon (1979); Cowell (2000); Conceicao and Ferreira (2000);Elbers et a1 (2005). 130The measure used here is Theil L or mean log deviation. This belongs to the general entropy class of inequality measures, with the aversionparameter to inequality among the poor being set to zero. inequality accounted for only about 17percent of total inequality in ~ ~ 0 0In.short, the bulk of the total ' ~ ~ interpersonal inequality is explainedby within-District inequality.132 Table 6.2: Between and Within-Group Decomposition of Theil Index of Inequality at District-Level Level of Decomposition:64-District 1991192 199912000 Number of Sub-Groups 61 63 Within-Group Inequality 0.11(85%) 0.16 (83%) Between-Group Inequality 0.02 (15%) 0.03 (17%) Total Inequality 0.13 (100%) 0.19 (100%) Source: HIES and Staff calculations Note: Figures in parentheses represent percentage contribution of each component of inequality to total inequality. 111. Correlates of Sub-National Expenditure Dynamics 6.15 In this section we test for the correlations between cross-District per-capita expenditure growth over FY92-05 and available policy and institutional variables. The choice of the variables is essentially dictated by data availability - for instance, a notable omission is a proxy for saving/investrnent across the Districts as the relevant data are not available. District level GDP data are not used since they coverjust four years -FY1996-2000. Key Bivariate Relationships Figure 6.4: Initial Expenditure Level and Subsequent Expenditure Growth 6.16 Districts with lower per-capita expenditure in 8 FY92 experiencedfaster subsequent growth inper-capita CV R sq. = -0.49 expenditure. This is consistent with the predictions of 0 0 standard neo-classical growth models, as per which 2" 4 - o o regions would converge toward a (more-or-less) common ; :: 2 - steady-state under conditions of similar production a.S functions and economic conditions shaping growth. In O - t .- B this case, Districts that are further away from the steady -2 - a oo state, i.e., the poorer regions, would grow faster by 2 -4- accumulating faster the factors of production, and vice 2 O versa for the richer regions. A pattern of mild -6 , I 8.8 9.0 9.2 9.4 9.6 9.8 10.0 unconditional convergence in expenditure level may be seen in Figure 6.4. Perhaps, the lack of ethnic Real Per Capita Exp in 1992 (natural log) fragmentation along caste-ethnicity lines combined with Source: HIES and Staff calcularions high population density fostered faster "geographic diffusion", especially in terms of informationhowledge diffusion. There are many such examples of fairly rapid spread of new "technology" such as from Green evolution,'^^ fertility control,'34 and microfinance group formation, which could have contributed to this observed ~0nver~ence.I~~ Homogeneity of population also allows greater mobility of labor within the country, which, through remittances, fosters faster convergence in expenditures across geographical units. It is important to note, 131 This is similar to what has been reported for some of the African and Latin American countries such as Madagascar, Mozambique and Ecuador (in all these countries the matched contribution was restricted to 20-25%) but much lower than other larger economies such as Russia (33%). See Kanbur and Venables (2005). 13' Analysis of inequality in interpersonal income per se goes beyond the scope of the present paper. Factors contributing to rising inequality as seen through the prism of income source-decomposition analysis has been discussed elsewhere (see, for instance, Khan and Sen (2001); Khan and Sen (2004)). '33Ahrned et a1(2000). 134Arnin and Basu (2000), and Dev et a1 (2004). 135Morduch and Aghion (2005). however, that the convergencerate indicated by the relationship in Figure 6.4 is very slow: it would take close to 150 years for the Districts to converge fully to a single level of per-capita expenditure at the indicated pace. Figure 6.5: Initial Gender Inequality Measures 6.17 Districts with higher gender inequality 2 and Expenditure Growth over 1992-2000 , experienced slower growth. Conceptually, gender R (sq.) = -0.38 - inequality can affect growth through two principal g 0 0 transmission mechanisms. With higher gender 4 - inequality, especially in human capital, there is less 5 -. - likelihood of female voice and female agency, 5 leading to high fertility and high population growth 0 - regime, which in turn, has adverse impact on the 2 -2 - 0 0 0 0 subsequent per capita income g r 0 ~ t h . l ~The ~ 5 second channel is through malnourished mothers - -4 - Source:HIES and Staffcalculations J (resulting from gender biases) having malnourished -6 children, who then tend to have poor schooling 2'0 i l o 4'0 6'0 8'0 1 6 0 i;o i;o performance, which can lead to lower future M a l e a d v a n t a g e in adult literacy rate ( 1 9 9 1 ) productivity, thus depressing the rate of long-term growth.13' AS a proxy for gender inequality we use initial male advantage in adult literacy rates.138 Gender inequality measured as such is seen to be negatively and significantly correlated with subsequent growth in per-capita expenditure (Figure 6.5). One implication of this is that the recent progress in reducing gender inequality in schooling (through incentive schemes) and in child mortality holds promise for long-term growth in Bangladesh. 6.18 Initial inequality in landownership is ,.-. Figure 6.6: Large-farm Landholding and Growth --I negatively correlated with Feudalism in land R (sq.) = -0.30 relations has been identified as a particularly sclerotic factor negatively influencing investment in new technology in agriculture and slowing down the pace of O broad-based human devel~~ment.'~~"~'Bangladesh, however, saw semi-feudalism only in patches: in the traditional Jotdari belt of North Bengal corresponding to upper Rangpur and Dinajpur districts in the Rajshahi Source: HIES, 1983/84Agriculture Census division; in the haor areas of the Sunamganjbasin of the and Staff calculations Sylhet division; and in some parts of the South where -6 . O O .04 .O 8 .1 2 .1 6 "reverse tebhaga" prevailed.'42 Even so, as seen in Area under large farms. 1983-84 Figure 6.6, "share of cultivated land under large- 136See, Eastwood and Lipton (1998). 137See, Glewee and King (2003); Osrnani and Sen (2004). 13'This is measured as 100*(maleadult literacy rate in 1991-female adult literacy rate in 199l)/(female adult literacy rate in 1991). 139See, Alesina and Rodrik (1994); Deininger and Square (1996); World Development (2005). 140See, Sobhan (1992) for a cross-countrysurvey. 14' There is a large-body of old-generation "mode of production" and new-generation "growth" literature on this subject. World Development Report, World bank, 2006, provides a useful survey on why high land inequality matters. See, Easterly 2000 on the long-term impact of feudalism on growth in Pakistan; Bardhan 1996 on the general importance of land-reform for faster progress in agricultural technology and human development; Dreze and Sen (2000) on contrastingregional growth experiences in the context of India. 14'"Reverse tebhaga" corresponds to the system where tenants are supposed to give two-third share of the produce to the landlord although the costs of production were entirely borne by them. This has become much less common these days even in the once-prevalent Southern pockets. landholding" (i.e. with land-size greater than 7.5 acres according to the 1983184Agricultural Census) is negatively correlatedwith expenditure growth at the District level. 6.19 Initial expenditure inequality is negatively " z Figure 6.7: Initial Inequalityand Growth associated with expenditure growth. The literature is R (sq.) = -0.36 increasingly looking at the possible adverse influence of 6 - high initial income (expenditure) inequality on the rate 0 0 .. - of subsequent growth as well as on the pace of future I lo 3 O poverty reduction.'43 Initial inequality can reduce i - growth by adversely affecting investment in human - 80"B capital. Initial inequality can also increase credit .-," O O - O 0 constraintsby reducing access to collateralizableasssets, -2 thus adversely affecting potential invesbnent. Further, - -4 - high initial inequality may lead to low poverty- Source:HIES hd StaffCalculations responsiveness of growth creating social unrest or 2 ,Go -6 ,d5 ,yo ,io , ,25 political tensions, which in turn, may generate Initial expenditure inequality, 1991 investment uncertainty, depressing growth. In some circumstances, especially in democracies (though not typically an empirical regularity in the context of developing countries), high initial inequality can lead to electoral pressures "from below" on the part of have-nots for additional poverty-transfers likely to be financed through greater distortionary taxation on the rich investors. This, in turn, can have negative impact on productive investment and growth. The inequality measure used in this exercise relates to "mean log deviation" (see, Annex Table 6.3). Figure 6.7 shows a significant negative relationship between initial expenditure inequality and subsequent growth. Sub-NationalExpenditureDynamics: Results of the Multivariate Analysis 6.20 Each of the variables noted above is also significantly correlated with per-capita expenditure growth in a multivariate regression analysis - with per-capita expenditure growth as the dependent variable and the rest as explanatory variables (Annex Table 6.2). As predicted, and consistent with the bivariate relationships seen above, expenditure growth is negatively and significantly correlated with initial levels of expenditure, gender inequality, asset inequality, and income inequality. In addition, we also include in the regression access to electricity and degree of urbanization in 1991 as explanatory variables, although in separate regressions since the two are highly correlated. The coeffiientsfor both access to electricity and degree of urbanization are positively and significantly correlated (at the 99 percent lever) with expenditure growth in the multivariate regression. It is worth noting that both variables vary considerably across Districts. In 1991, access to electricity varied from just 2 percent in the Kurigram district to 69 percent in the Dhaka district. Similarly, the degree of urbanization varied from only 7 percent in the Gopalganj district to 88 percent in the Dhaka district. 6.21 Resultsfurther indicate thepresence of "conditional"regional convergence: i.e., Districts with lower initial per capita expendituretended to have higher growth in expenditure per capita. However, the speed of convergence is quite slow- the estimated time for full convergence in per-capita expenditure is about 90 years. 143Ray (1998) surveys the literature on the effect of inequality on growth. Ravallion (1997) and (2000) discusses impact of inequalityon poverty reduction. 77 IV. Discussionof Results and Implications for Policy 6.22 This chapter highlightedthe considerable(althoughnot unusual in any sense) spatialvariabilityin GDP and expenditure levels and in expenditure growth in Bangladesh. Results presented in the chapter indicate that the level of inequality between Districts accounts for less than one-fifth of the total inequality in the country: interpersonal inequality accounts is the main contributor to that. The chapter also showed expenditure growth at the District level to be negatively correlated with initial levels of gender disparity, expenditure inequality,and asset inequality, and positively correlated with initial access to electricityand degree of urbanization. It further showed some evidence that Districts with lower initial per capita expenditure, in general, appear to have grown somewhat faster, indicatingmild convergence in expenditure per capita among the Districts. Any convergence among Districts is quite slow, however. Unconditional convergence - the rate at which convergence would take place without addressing the major policy and institutional differences among Districts - could take more than 150 years. Policy support for the laggingregions, therefore, is clearly quite important for shared growth within the country. 6.23 Is there, then, a rolefor "regionalpolicy" that can help accelerate convergence among sub- national units? If so, what forms it should take? Unfortunately, analytical limitations at this stage prevent us from making a clear leap from identifying factors that are correlated with and possibly impact regional growth to elaborating clear policy choices that would address growth constraints at the regional level. Sub-national income dynamics are also shrouded in economic and political complexitiesthat need to be further explored. Particular considerationhas to be given to the information, resource, and capacity constraints w i t h which policy makers operate. Our emphasis at this stage was simply to highlight the differencesin sub-nationalincome and expendituredynamicsand provide a preliminary assessmentof the possible causes, in order to alsobetter understand the growth process at the national level. Chapter 7: Governance and Growth: the Bangladeshi I. Introduction 7.1 Bangladesh's impressiverecord of economic growth and social change has been achieved despite weak governance, particularly high levels of corruption and a perceived worsening of regulatory quality. Bangladesh is thus an outlier in cross-country studies which relate governance to economic growth. What explains this relatively strong performance despite apparently poor governance? And would this performance be better and more sustainable if governance were better? This chapter attempts to grapple with these questions. Although there are no clear answers, the cost to growth and social development from poor governance is clear enough. 7.2 The chapter begins with a brief overview of whether governance matters to growth. It summarizes what we know about the quality of governance in Bangladesh. Governance is then unbundled in an effort to demonstrate that Bangladesh's governance also has some positive aspects to it although these often not recognized by observers. The chapter then focuses on the negative aspects that adverselyaffect growth. It concludes by arguing that building on its many successesand addressing these negative governance factors could help remove some of the constraints that are preventing Bangladesh's growth performance coming closerto its faster growing neighbors. II. Does GovernanceMatter to Growth? 7.3 Research over the last decade has shown increasingly compelling evidence that good governance matters to long-run growth. This research has included economic history studies of development over the last millennium as well as cross-country studies that have tested the statistical relationship between governance and growth.'45 One study shows that absolutist governments were associated with slow growth of cities in the 800 years preceding the industrialrev01ution.l~~Large cross- country studies have shown that political instability adversely affects property rights,14' and that subjective assessments of political risk to overseas investors provided by commercial firms are closely associatedwith lower investmentand growthrates.148 7.4 These statistical relationships do not say much about the direction of causality. Governance may have simply improved as income levels increased More recent studies have attempted to address this weakness. A recent comprehensive study asks what the independent contribution of geography, integration and institutions is to the cross-national variation in income level, and concludes that institutions trump everything else. Once institutions are controlled for, global integration has no direct effects on incomes, while geography has at best weak direct effects.14' A recent World Bank staff report '"This chapter draws on a discussion of the Governance conundrum in the recent World Bank Group's "Country Assistance Strategy for the People's Republic of Bangladesh for the Period FY-06-09", Report No. 35193 dated February 6,2006,WashingtonD.C. 145Governanceis broadly defined in these studies to mean the exerciseof power through the country's economic, social and political institutions. These include the processes by which govemments are chosen and replaced, held accountableand monitored, the capacity of govemment to formulate and implementpolicies and manage resources efficiently,and the respect with which citizenshold the institutionsthat govern social and economic interactions among them. 146De Long, J. Bradford and A. Shleifer (1993). 14'Barro, Robert (1991). 14'Knack, S., and P. Keefer (1995). 149Rodrik, D., Subramanian,A., and F. Trebbi (2002). also attempts to address the causality issue by measuring long-term growth during periods entirely subsequent to when the quality of governance assessments were made (Figure 7.1). The strong relationship between governance and growth remains, and is not very sensitive to whether developed countries are included in the sample or not, although it becomes slightly weaker if countries in the East Asia and Pacific Region are excluded. Figure 7.1: Governance and Growth = Governanceand Growth, 1982-2002 ..! - V) CHN e! -06 N - I .- ZAR -.. ICRG index (residual) Note: The scatterplot in Figure 7.1 shows the partial relationship between the quality of governance in 1982 and income growth in the subsequent two decades (controlling for schooling levels and initial income levels) for all countries for which data was available except 22 developed countries. (Global Monitoring Report, World Bank, 2006). 7.5 The conclusions reached by these studies are not uncontested. The studies rely heavily on subjective indicators, since objective indicators that measure constrained behavior are almost by definition impossible to find. And objective indicators that measure intent but not practice are meaningless.lsO It is also argued that subjective indicators are prone to ideological biases and overgeneralizations on the part of businessmen surveyed. The effects of recent incidents may overly affect judgments by citizens and businessmen. And those surveyed are often part of a small group of "country watchers" who tend to talk to each other and are influenced by each other. Moreover, respondents to such surveys may confuse cause and effect. A strong growth performance may lead some to assume strong underlying governance (as with East Asia during the 1980s and early 1990s), while ignorance about strong development gains in countries that are not the focus of much international attention except during times of crises may lead to the opposite conclusion (e.g., Bangladesh). Moreover, indicators typically used, such as the protection of property rights, may be unduly influenced by non- institutional factorssuch as the emergenceof civil unrest. 7.6 However, there is steadyprogress in improving the quality of governance data andparticularly in the design of questionnaires eliciting the views of businessmen and citizens to reduce the noise in the data. Moreover, the users of this data are well aware of these issues and have been taking increasing pain to deal with the underlying concerns through a range of statistical techniques.lsl Above all, perceptions become the reality, influencing as they do behavior and, particularly, investment decisions of foreigners,and thereforeneed to be taken seriously. 150See Weyland, Kurt (2003). 15'See Kaufmann, Daniel, Kraay, Aart and Mastruzzi, Massimo (2006). 7.7 That we have evidence that governance matters to growth however tells us little about why it matters and how it influences growth. The cross-country studies have failed to establish a causal link between particular forms of institutions and economic growth. As academic Dani Rodrik observes, "we know that growth happens when investors feel secure,but we have no idea what specific institutionalblue prints will make them feel more secure in a given context".152Certainty of property rights is clearly important to investors, but this implies nothing about the form that property rights should take.Is3 Problems of identifying appropriatepolicy prescriptions are accentuated by less than full transparency in the criteria, information sources and methodology used by risk assessment f m s . For example, it is possible that countries are only downgraded for violations of property rights that the risk assessors deem to be harmful to growth. If so, an exclusive emphasis on the security of (all) property rights might actually reduce growth in the (admittedly few) developing countries with growth-oriented high-capacity governmentsthat might otherwisere-allocaterights in ways that are favorableto long-run growth.154, 155 7.8 Understanding individual country contexts and how institutions work in particular historical and cultural settings is important to understanding better the governance-growth nexus. This chapter is an attempt to improve our understanding of this relationship in the particular context of Bangladesh, where the economyhas seen an impressivegrowth performance despite poor governance. 111. Bangladesh's Mixed Record of Governance 7.9 Bangladesh's strong growth and macro-economicperformance highlighted in VolumeI of thk report contrasts sharply with the relativelypoorperceptions of governance in the country. In the most recent governance data set released by the World Bank Institute (WBI) for 2005,lS6Bangladesh scores poorly on all six indicators, ranking between the bottom seventh to tho-second percentile among some 210 countries (Figure 7.2). Bangladesh's percentile rankings were, respectively, 6.6 for political stability, 14.9 for regulatory quality, 19.8 for rule of law, 7.9 for control of corruption, 21.1 for government effectiveness, and 31.4 for voice and accountability. Within the South Asia regon, with at a confidence level of least 90 percent, Bangladesh fared worse than India on each of the six indicators and Sri Lanka on rule of law and control of corruption, and better than Nepal and Pakistan on voice and accountability. 7.10 A comparison of thepoint estimatesfor these indicators over time suggests that Bangladesh's performance has worsened on all six indicators between 1998 and 2005. However, because of large, though declining, margins of error, there is not enough evidence to conclude that Bangladesh's governancehas become better or worse, except in the case ofpolitical stability, where it is possible to say at a 90 percent confidence interval that there was a perceived decline. On both control of corruption and regulatory quality,there is at least 70 percent chance that there was deteriorationover time Is2Rodrik, Dani (2006). lS3Rodrik, et a1(2002), cite the experience of China and Russia: China retains a socialistlegal system, while Russia has a formal property rights regime in place. Yet Russia scores lower than China in property rights indicators. Credibly signaling that property rights matters is more important than having formal property rights on paper, they conclude. Is4Khan, M (2002). Is'For example, land reforms in Taiwan and Korea are often credited in part for their subsequent economic success. Is6The increased awareness of the importance of institutions to development outcomes has been accompanied by a large body of work to measure the quality of governance in different countries. The WBI has developed a set of indicatorsaimed at capturing these broad dimensions of governance. These indicators aggregate a large number of surveys of businesses, citizens and experts both within and outside countries to measure performances in six key dimensions of governance:voice and accountability, political stability, government effectiveness, regulatoryquality, rule of law and controlof corruption. Figure 7.2: Bangladesh Governance Indicators 1998 and 2004 BAN(XFIDESH (2005) Y o l c ~ad kc-tabilitg P o l i t i d StabilitdNo Yi* 6-rnmt Errcsllvenesr Rwlatarg wt!, me LSU h t r o l of CorruptLm Source: Kaufmann D.,A. Kraay, and M. Mastruzzi 2006: GovernanceMatters V: GovernanceIndicatorsfor 1996-2005. 7.11 While these data rely on perception-based surveys, the storyline that emerges is reinforced by the data emerging from surveys on Investment Climate (done jointly by the World Bank and the Bangladesh Enterprise Institute) and Doing Business (done by the World Bank) that rely on opinions based of those actually engaged in business in Bangladesh. While subjective bias is likely even here, the information collected alsorecords specific data on costs associated with governance. Investment climate survey data from 2002 (Table 7.1) show the bureaucracy delivering in some areas and not in others. Senior management time spent dealing with requirements of regulations is a modest 4 percent, just over half the regional average, while enterprises on average find a much higher level of consistency of oficials ' intelpretation of regulations than in South Asia or all countries surveyed. Yet, Bangladesh enterprises report almost twice the level of concern with economic and regulatory policy uncertainty than in South Asia as a whole. Similarly tax administration was seen as a major or severe obstacle by half the firms surveyed, double the global average and more than double the South Asian average, while customs and trade regulations also received poor ratings with 42 percent of firms seeing this as a major or severe obstacle. Corruption is clearly a major concern. Some 58 percent see corruption as a major obstacle, as against a global average for the survey of 30 percent and a South Asia region average of 25 percent. Corruption is thus the second most important source of concern for investors in Bangladesh. Some 86 percent of firms expect to give gifts in meetings with tax inspectors, almost twice the level for South Asia as a whole Lack of confidence in the judiciary among Bangladeshi enterprises at 83 percent of the firms surveyed was the lowest rating among all countries surveyed. Crime, theft and disorder were seen as a major or severe obstacle by 39 percent of firms, again substantially higher than in the region as a whole (15 percent) and globally (20 percent), although the actual cost of such losses was around the South Asia average (0.6 percent of sales). Table 7.1: InvestmentClimateSuwey Indicator IBangladesh I South I All 7.12 Information fiom the World Bank's Doing Business 2006 database is also not encouraging (Table 7.2) overall. There are more procedures involved and it takes more time to start a business in China than in Bangladesh,and yet the associated cost in China is half that in Bangladesh. Once firms are established they are subject to frequent visits fiom a variety of government agencies (some 17 a year on average in 2003 compared to some 7 a year in India in 2004). The number of procedures and their time requirement for registering a property is among the highest in the world - it takes less than one-fifth as much time to register a property in the median LIC as in Bangladesh. Bangladesh does a bit better on contract enforcement, where the number of procedures, time, and cost involved are each lower than in India, Palustan and the median LIC. Exporters need a lot of signatures, which take a long time (35 days on average, compared with 20 days in China) and undercuts competitiveness of Bangladeshi exporters. Recovering a loan in a bankruptcy can take up to 4 years and cost up to 8 percent the value of the estate, with the recovery rate a low 24 cents on the dollar. Table7.2 Cost o 'Startinga Businessin Bangladeshand Comparator Countries Startinga Business Registeringa Property [ EnforcingContracts I ExportRocedure Procedures I Time 1 Cost Procedures1 Time 1 Procedures1 Time 1 Cost IDocumentsI Signatures China India Malaysia Pakistan Sli Lanka Thailand 6 71 34 390 31 8 13 m e : Doing Business Indicators 2006, World Bank. IV. Unbundling Governance in Bangladesh: Some Strengths 7.13 How has Bangladesh enjoyed such relatively strong economic outcomes with such poor governance? The mixed record of governance shown above suggests however that Bangladesh has enjoyed governance successes in some key areas. There are at least five areas where Bangladesh has unquestionablydemonstrateda capacity for good governance. 7.14 First, the state created spacefor the emergence of what is a vibrant domestic private sector. It has done this in part by increasingly introducing macroeconomic discipline, as evident fiom low inflation rates, fiscal deficits and external indebtedness, thus reducing the crowding out of private sector investment. One aspect of the enhanced macroeconomic discipline is the improved fiduciary regulation of commercial banks by Bangladesh Bank. The Government has also had some success in reducing the non-performing loans of nationalized commercial banks (NCBs), a major source of weak governance, by capping new lending to private borrowers, enforcing a ban on loans to defaulters, closing 100 bank branches. These measures have resulted in a sharp fall in the share of the NCBs in total bank lending; from 70 percent in 2000 to under 40 percent today (see Chapter 8 for details). In addition, the state has opened up the economy through trade liberalization, although, as argued in Chapter 5, there remains a strong anti-export bias confronting Bangladeshi producers. Nevertheless, from a long term perspective, trade liberalization and the successful encouragement of garments exports have led to a near doubling in the share of trade in goods in GDP since 1990,contributing to an enlarged role for the private sector. 7.15 Second, successive governments have encouraged the migration of Bangladeshi workers to West Asia and other destinations, and created a domestic economic environment that has encouraged these workers to remit large sums of money to Bangladesh. These remittances, which amount to over US$6 billion a year (including those through informal channels), have in turn stimulated domestic demand in the country side, and financed growth inducing investments in land, agricultural inputs, human capital and business start-ups. 7.16 Third, Bangladeshi governments have been relatively good at making wisepublic expenditure choices. Military spending has been kept under check, and at 1.1 per cent of GDP, is less than half the regional average, leaving fiscal space for pro-poor spending. It has also financed rural inhstructure, particularly roads connecting about 1,400 of 2,100 growth centers and markets and thus also contributing to the significant increase in non-farm incomes. Government budgets have also ensured a sustained if still modest injection of public funds for health and education that have increased the quality of human capital and contributed to the significant strides in attaining the Millennium Development Goals (MDGs). 7.17 Fourth, the state has recognized its limitations in meeting the strong demand for public services and has created space and forged partnerships with NGOs and the private sector to help deliver social services. Unlike many developing countries where the state crowds out civil society or co- opts NGOs, Bangladesh has seen the emergence of effective partnerships between large, effective NGOs and the state to deliver micro-credit and to provide education and health services. Bangladeshi NGOs account for just under a tenth of health and education spending. But they have also emerged as major social enterprises with considerable capacity to deliver a range of developmental services that have contributed to both growth and human development. The state has also pragmatically encouraged the private sector to help provide secondary and tertiary education. 7.18 Fifih, the state has greatly improved its capacity to manage natural disasters. The macro- economic impact of natural disasters has been considerably reduced. The 1998 flood for instance was far more severe than that of 1988. But per-capita GDP growth stood at 3 percent in FY99, whereas a decade earlier per-capita GDP growth was negative in the flood year.'57 The factors contributing to this improvement include the reduced dependence on farm income, early warning systems, the construction of cyclone shelters, and better relief efforts including reduced leakages in food distribution, allowing private sector imports of food, and well targeted Voluntary Group Feeding systems. The response capability of civil administrations has improved markedly and public campaigns to educate households on food and water safety precautions during floods and cyclones have proved effective. NGOs play a major role in disaster prevention, recovery and relief, and an alert opposition and free press have motivated successive governments to respond promptly to disasters. 7.19 What is clearfrom the above is that the State has played a signifcant role in the growth that Bangladesh has enjoyed. It has done that by not crowding out the private sector and civil society, by delivering with some effectiveness key services, such as macroeconomic management, human '57See Sen,B. (2004). development and disaster prevention, and providing a certain degree of continuity in govemment policies despite changes in government. These positive governance aspects arise from recognition by the state of its weak implementation capacity and stepping back from its role as a producer to that of regulator and facilitator. While the stepping back has contributed to a strong private sector response and has been crucial in the delivery of essential services where NGOs and the private sector have stepped in, the government's capacity to regulate and facilitate is uneven and governance issues have been a serious impediment to delivering on these new roles. 7.20 However, these positive governance aspects are somewhat overwhelmed by the significant weaknesses in a number of other governance areas. The weaknesses lie principally in the country's core governance systems: its politics, the way it h d s elections, the way it manages its public finances including revenues and expenditures,the way it delivers public services and the legal andjudicial services it provides its citizens, all of which have a profound impact on development and economic growth. There are serious sectoral governance challenges as well which have an adverse impact on growth. These are discussed in Chapters 8 for the financial sector and in Chapter 9 for the infrastructure sector. V. Challengesin Core Governance 7.21 The most serious and difficult governance challenge lies in the political domain where confrontational politics and the high cost of election finance have severe adverse economic consequences. Governance challenges also severely impede public financial management. There are major shortcomings in tax administration (poor taxpayer services, lack of transparency in collection, inadequate audit and enforcement, and protracted taxpayer disputes) and the resulting low revenue collection creates a chain of disincentives to good governance, from low salaries to civil servants to inadequate operations and maintenance expenditures and lack of effective checks and balances. Financial accountability is weak, with flawed procurement processes, poor financial controls and inadequate external checks and audits. Essential infrastructure is not properly maintained for lack of resources and leakage, while the state has little or no capacity to finance infrastructureinvestments. This in turnleads to a reliance on private provision of infrastructure, which is poorly managed and regulated by the state resulting in a serious infrastructure crisis. Bangladeshis in general view the civil service as being of low quality, which further feeds the vicious cycle of poor governance. Finally, thejustice sector has also been a source of weakness. Although the Supreme Court and the seniorjudiciary are exceptions to this rule, the poor state of the lowerjudiciary, the police and the prosecution services means that the citizen's first encounter with the justice sector tends to be a disappointing one. Above all, the courts have difficulty enforcing contracts and enforcing the rule of law, two essentials to creating an enabling environment for the private sector. These issues are discussed in more detail below. Political comvetition 7.22 Political power has been concentrated in two major parties - the Awami League and the Bangladesh Nationalist Party - that have dominated Bangladesh's politics since independence. A first-past-the-post system of elections results in large stable majorities for governments. A constitutional amendment that forbids members of political parties from voting against their parties in Parliament reduces the importance of Parliament as a check on the executive. The lack of inner party democracy prevents these parties from being held accountable by their grassroots. The strong centralization of the state in Bangladesh precludes the sharing of effective political power through local governments. 7.23 The outcome is a system in which the winner takes all in elections and the loser has difficulty reconciling to the loss. Bitter relations between the two parties lead to frequent confrontations, which oAen shift to the streets. Frequent hartals, or work stoppages, are one consequence, which according to a UNDP report costs the economy some 3-4 per cent of GDP annually. The dependence of political parties on muscle power results in the criminalization of politics and increasing political violence. Fundamentalistgroups which appear to have a rather modest public followinghave neverthelessexploited this dissension among the secularparties in recent years to pursue their own agenda. 7.24 Bangladesh is not unique in its confrontationalpolitics; developed and developing nations have shown this tendency. But thefractiousness contributestopolitical uncertaintyand toperceptions of political instabilityfor outside investors who may not always understand that much of this tends to be "soundandfury,"signifling little. Potential foreign investors taking a long-term view of investments in developing countries tend to stay sway from what they see as a high risk environment which has a significantimpact on growth. The answer lies in a number of political reforms that improve the quality of political competition in Bangladesh, including strengthening the role of Parliament and parliamentary committees including ensuring an effective role for the opposition in the work of parliament, decentralization of power to local governments, requiring inner party democracy for political parties contesting elections and strengthening civil society watch dogs of the political process. These changes can only come if the people of Bangladesh demand it and cannot be imposed from outside. The high cost of elections 7.25 Elections in Bangladesh, like in many democracies, tend to be a costly affair, and certainly cost much more than the Tk 500,000 that election rules permit for each parliamentary candidate. Parties and individuals wanting to get elected must raisefunds tofight elections and this opens the door to undue influence of money on public policy. Moreover, because actual expenditures on elections exceed prescribed limits they result in a lack of transparency about the source of such funds and blur the line between a parliamentarian's public and private use of such funds. State capture resulting from the incurring of debt to vested interests distorts trade and industrial policies and erodes competition. Public funding of elections, reducing the cost of fighting elections through limited free access to TV and radio for candidates,and requiring transparencyrelating to donations for elections will help. Enforcingexisting lawsrequiring candidatesto declare their assets and liabilitieswill be critical. Low revenue mobilization 7.26 Bangladesh's revenue efJbrt has been weak, and for the past several years it has stagnated. With a population of 140 million, the country has only 1.4 million registered tax payers and 300 businessesregistered for its value added tax. Total tax revenue is a low 8.5 percent of GDP. This reflects the low elasticity of revenues with respect to income growth due to a heavy dependence on import taxes (45 percent of total tax collections) and numerous exemptions, tax holidays and preferences. It also reflects a tax administration that is weak and open to corruption. Investment climate surveys show that half the firms see tax administration as a major or severe obstacle to investment, almost double the average for all countries surveyed, and 42 percent of firms see customs administration in the same way, more than double the average for all countries surveyed. It takes as much as 11 days for imports to clear customs, nearly twice as long as the average country surveyed. While customs has seen some improvement of late, shippers still face cumbersome processes including large number of approval processes (15 signatures for exports on average and 38 for imports) and many opportunities for face to face contacts with customs and tax inspectors and staff which create opportunities for rent seeking. Tax administration is complicated by a cumbersome management system that lacks clear direction and leadership and an organizationalstructure based on type of tax which tends to duplicate functions (such as audits and collection) and undermine the exchange of information. This forces taxpayers to deal with different units on similar issues, with little scope for redress. 7.27 Apart from the direct disincentive effectsfor potential investors, governance issues in tax and customs administration raise the cost of doing business in Bangladesh and erode the country's competitiveness. They also reduce the resources available to the state to pay adequate salaries to civil servants, meet operations and maintenance (O&M) expenditures, and finance much needed infrastructure for growth. Low provisions for salaries and wages and O&M feed the vicious cycle of corruption which constitutes an acrossthe board tax on growth. 7.28 The National Board of Revenue (NBR) is embarking on a reform program to address these issues. The government has established Large Taxpayer Units for Income Tax and VAT to help improve tax collection and reduce the costs of dealing with the NBR. A strategic development plan has been recently approved aimed at raising revenue collections, organizing the NBR along functional lines, and ensuring that all businesses are treated uniformly and equitably. 7.29 A comprehensive tax administration modernization program is being developed by GOB. The modernization program will include developing new functional areas such as tax payer services, audit functions to detect underreported income and collection enforcement. These new functions will be piloted by the Large Taxpayer Units for income tax and VAT, and will be supported by reengineering processes and automation. The NBR's composition will be strengthened and based on functional lines. Human resource policies and procedures will also be improved to ensure more professional and competitive recruitment and promotion policies, training, compensation and career development. The NBR will also be encouragedto work with exporter and taxpayer associationsto ensure that the quality of service improves and to make the NBR more client-hendly. Periodic surveyors of taxpayers, exporters and the general public will monitor perceptions of the NBR's progress in addressing the concerns of its clients. Public financialaccountabilitv 7.30 A recent World Bank review of institutional arrangements for public expenditure,financial management and procurement revealed the need to significantly strengthen Bangladesh's public financial management. Key links in the chain of public financial accountabilityare flawed: A serious weakness is the lack of a strategic planning process reflected in the routine over- programming of the annual developmentplan. This results in the reliance on discretion rather than rules, creates opportunities for rent-seeking and leads to too many under funded projects, implementation delays and chronic under expenditure. Budget planning is fragmented between recurrent and investment budgets and preoccupied with inputs rather than outputs and outcomes, resulting in weak performance orientation. All these contribute to weakening the impact of public investment on growth and development. Weaknesses in budget execution related to poor accounting practices, including reliance on manual systems which weaken accountability and transparency, and reporting delays. All these contribute to leakagesreducing the effectiveness of budgetary spending. State-ownedenterprises do not use internationalaccounting standards. This contributes inter alia to a lack of transparency about contingent liabilities in the budget. Weakinternal and external controls is reflected in the lack of accountinginformation and proper financial controls in line ministries, delays in the publication of audited reports, unsatisfactory quality of the external audit function and ineffective parliamentary oversight over the entire budget process. A new Anti-Conuption Commissionhas so far not been able to make an impact as an additionalweapon in the government's armory against corruption. 7.31 The Government has begun to address some of these weaknesses. It has strengthened strategic budget planning by expanding the medium-term budget framework (MTBF) approach. The MTBF approach has been introduced on a pilot basis in ten ministries (6 in FY06 and 4 in FY07), and there are plans to cover another four ministries in the FY08 budget. This will bring MTBF coverage to more than 65 percent of total budgetary expenditure. The emphasis at this initial stage has been on better integration of the revenue (recurrent) and development (investment) budget planning processes at the center and line ministries, by unifying the budget preparation process and issuing an integrated budget ceiling for line ministries. The MTBF approach has also given more flexibility to the line ministries in their spending allocation decisions, and has encouraged the substantive involvement of senior officials in the budget formulation process, which otherwise was considered routine work and typically undertaken by lower- level officials. Despite such progress, the process is still in its infancy, and needs to be significantly deepened before it can produce tangible gains in linking policies and budgets and contribute to better budget and service delivery outcomes. 7.32 To strengthen budget execution, serious efforts are being made to improve the accuracy and timelines of expenditure and strengthen internal and external accountability, audit and scrutiny. A customized computerized accounting system has been rolled out 60 district accounts offices and 49 chief accounting offices. In early FY07, the Ministry of Finance and Planning issued a circular notifying the Government's decision to adopt the International Public Sector Accounting Standards (IPSAS) cash standards for accounting. Several government agencies have started implementing this circular. The Government has also reduced a major conflict of interest by separating out the internal and external audit functions. To strengthen the internal audit function, the Government has prepared a Public Expenditure Management Manual and the Internal Control Manual, and taken steps consistent with the manuals, including separation of cash management from debt management. The time lag for final audited accounts has been reduced from two to one year. The Comptroller and Auditor General (C&AG) has introduced performance and entity based audits on a pilot basis and is planning to roll them out across various government units. Audit committees are being formed in ministries, and the Government remains committed to malung the C&AG functionally and administratively independent 7.33 Reform eflorts have also sought to strengthen the transparency of fiscal information and key institutions of accountability in public financial management. To this end, GOBdeveloped baseline ratings of the core PFM functions in early 2006 and more recently prepared a comprehensive, medium- term, rolling Public Financial Management Improvement Plan (PFMIP) that lays out a series of second- generation PFM reforms. Among the initial implementation measures, the FY07 Budget Summary, for the first time, included an explanatory note that listed all explicit and counter guarantees provided by the Government against liabilities of SOEs; these amounted to US$2.5 billion (4 percent of GDP). Furthermore, in FY07, GOBbegan setting up a technical support unit for the Public Accounts Committee (PAC) in the Legislature and developing a system of information that would increase the transparency of its decision-making, once the elected Parliament returns. The PAC had made good progress in reducing the backlog of audited accounts for discussion, and its deliberations were beginning to be followed by the press in Dhaka, until the Parliamentwas dissolved in October 2006. Transparency and comvetitivenessof public procurement 7.34 A key element in strengthening public financial management is making public procurement much more competitive and transparent. Bangladesh, like many developing countries faces significant challenges in improving its public procurement, which typically has been subject to considerable delays, lacks transparency and is open to collusive practices. It is widely regarded as being the single most important source of leakage of public funds in Bangladesh. Reducing opportunities for corruption in procurement will significantlyenhance the effectiveness of public expenditures,reduce the costs of public infrastructure and donor financed projects and contributes to faster growth. Recognizing this, the Government has initiated a number of reforms in the area. Despite strong resistance from powerful vested interests, the previous Parliament adopted a new law on public procurement in July 2006, with many provisions that are in line with well-accepted international practices. The Government is now finalizing the rules to implement the law, and in the interim is using the regulations that were the precursor for the law. The regulations have introduced new standard bidding documents that significantly reduce the layers in the procurement approval processes. The Government has also put into place a unique procurementperformance tracking and monitoring system and is piloted it in a few agencies. 7.35 Moving ahead, strong compliance monitoring, not just by the Central Procurement Technical Unit, which manages the Government's procurement process, but also by civil society and NGOs through procurement watch activities will be crucial to effective implementation. Systematic outreach to potential bidders in the business community, including civil works contractors, consultants and suppliers, will be necessary to explain the new law and seek their cooperation in making it a success. Quality of the civil service 7.36 At the root of thefailure of the state to contributeeffectively to growth, as in many countries,is a civil service that is politicized, poorly paid and poorly managed. Bangladesh's civil service today is inadequately trained, hierarchical and secretive and subject to political influence. Lack of accountability contributes to poor service delivery, the deterioration in regulatory quality and many of the problems discussed above in public financial management and revenue mobilization. 7.37 Numerous commissions and reviews have addressed these concerns, and the Government has acted on some of the recommendations. The size of the public service and the wage bill has been successfully contained. A career planning wing of the Ministry of Establishment was created to focus on career planning and effective deployment of Class I officers. A training policy has been announced and training is now a criterion for promotion of officers. Key to further reform will be strengthening the personnel management system, ensuring a merit and performance based system of promotions toward which initial steps have been taken, a strategic use of postings and transfers to strengthen capacity and match available skills to requirements, a revised compensation package, particularly for the senior cadres of the civil service who are relatively poorly paid for their level of responsibilities. The rule of law 7.38 The rule of law is central to the growth process. While Bangladesh's Constitution has provided for all the formal requirements of a society based on the rule of law, in practice the system fails to deliver, as is evident from the low confidence in the judicial system cited above. With the possible exception of the Supreme Court, all the key pillars of the justice sector are weak, inefficient and prone to corruption. Law and order is perceived to be weak. The police are held in little confidence. In part this is a matter of staffing: its 80,000 complement makes the force quite the smallest in the sub-continent in relation to population. In part it reflects the emphasis placed on maintaining a modicum of public order and protecting the wealthy rather than preventing, detecting and investigating crime. A weak prosecution service and an inefficient and allegedly corrupt lower court system contribute to the lack ofjustice. 7.39 For ordinary citizens, access tojustice is impeded by the high costs of the system and the low probability of a successful resolution. For businesses, the inability to enforce contracts promptly and to secure property rights, particularly for land, greatly raises the cost of doing business in Bangladesh. Creditor and minority shareholder rights are well protected under the law, but enforcing these rights is another matter. But as is noted in Chapter 8 on the financial sector, lenders and borrowers must deal with a flawed collateral system. Unreliable land records, high fees and fraud hamper land titling. And as noted below, registering a property is cumbersome, and enforcing a contract through the courts is costly. Bankruptcies are not an easy solution to corporate distress. It takes 4 years on average to shut a business down, and you may recover 24 cents on the dollar, better than the regional average of 20 cents, but much lower than the 44 cents that are recovered in Pakistan and Thailand or even the 40 cents recovered in Uganda. While domestic entrepreneurs learn how to survive the system and absorb its high costs, foreign investors unfamiliar with Bangladesh and with plenty of other places to invest in are greatly deterred by these failures. 7.40 Comprehensive but well sequenced reforms of the justice sector are long overdue. The Government has begun modestly with a program to strengthen court administration and reducing delays through an automated case management information system, strengthened judicial training, upgrading facilities and other human resource needs in five pilot districts and the Supreme Court. Important steps have also been taken to separate the judiciary from the executive. DFID, Japan and UNDP are helping with police reforms. But more comprehensive reforms will be crucial if the failure to ensure the rule of law and protect property rights is not to prove a serious impediment to the acceleration in growth that is the topic of this report. Im~rovin~the regulatory environment 7.41 The formal hoops that new businesses must jump in Bangladesh may be relatively modest (Table 7.2), but the red tape is more formidable in practice than it is on paper. The high informal regulatory costs are especially wonying in the post-ATC environment where garments exporters must compete with low cost producers in China, India, and Vietnam. Adding even a few cents to the price of a garment can cause business to go elsewhere given the competitiveness of markets. The Government recognizes this and is working to streamline the regulatory process and strengthen its regulatory policy capacity. Industrial zoning programs are being considered to insulate investors from some of the constraints they face operating in Bangladesh. The Board of Investment, the Bangladesh Export Processing Zones Authority, the Registrar of Joint Stock Companies and other regulatory institutions are strengtheningtheir capacity to support investors and simplifying their procedures. M. Conclusion 7.42 Bangladesh has enjoyed good growth since the early 1990s despite weak governance. Growth has accelerated slightly in recent years, but as is argued elsewhere in this report, diversification of the economy and improving the environment for foreign direct investment will be central to bringing Bangladesh's performance closer to its neighbors in the region. Like Indonesia and other South East Asian countries in the early 1 9 9 0 Bangladesh is likely to soon discover that it is fast growing out of its ~ ~ current institutional environment. While governance may have been adequate to cope with an economy breaking out of stagnation and poverty, it is increasingly proving a bamer to the acceleration of growth that is needed to push Bangladesh firmly on the path of middle income status and poverty reduction. And yet, from a political economy perspective, the fact that growth has been adequate and rising reduces the pressure on political leaders to address the governance issue. 7.43 Addressing the governance challenge will also not be easy. The governance agenda is large and cuts across a wide range of institutions and threatens powerful vested interests. Developing a strategic, sequenced approach that relies on success in a few key areas to generate momentum and demand for reform in other areas will be crucial. Summoning the political will to do this will not be easy, and will itself depend on strong political leadership and a public that demands reforms and shows lower tolerance for weak governance. Chapter 8: Creating a More Efficient Financial System I. Introduction 8.1 Creation of a deeper and more efficient financial system can be considered a sine qua non for sustaining high growth and attaining the middle-income country status that is desirable for Bangladesh. It is reasonably clear that this will require a substantial change in the role of government - from an operator and arbiter in the financial system to a facilitator role. This implies not only divestment from the nationalized commercial banks (NCBs), a laudable reform process that is underway, but also de- politicization of the licensing process and a market-based framework for resolving bank failure. Most importantly, the governmentwould need to move away from the implicit guarantee provided to depositors and owners to applying the existing limited explicit deposit insurance for depositors, while simultaneously relying more on market participants to monitor and discipline banks instead of micro- managing financial institutions. This redefinition of government's role also applies to other segments of the financial system, such as capital markets and the micro-finance sector, and should be seen as an essential element in the governance reform agenda and in the movement from a relationship-based economy to a market- and arms-length economy. 11. The Structure and Recent Performance of the Banking Sectorin Bangladesh 8.2 Financial intermediation in Bangladesh, as in majority of the developing world, is heavily dominated by commercial banks. As of end-2005, the banking sector comprised four groups of "scheduled banks" (see Table 8.1): four NCBs, five government-owned development financial institutions (DFIs) (specialized banks that deal in "development financing"), thirty private commercial banks (PCBs), and nine foreign commercial banks (FCBs). While there are a number of investment banks, merchant banks, leasing and finance companies, insurance companies, and stockbrokers, these are small in comparison. As of December 2001, the banking sector accounted for 89 percent of the country's financial assets.'58 The story of the banking sector, therefore, is largely also the story of the financial sector in Bangladesh. Table 8.1: Structure of the Banking System (end-2005) BankTypes Number of Number of % of lndustry % of lndustry Banks Branches Assets Deposits NCBs 4 3386 37 41 DFls 5 1342 10 6 PCBs 30 1643 46 46 FCBs 9 41 7 7 Total 48 6412 100 100 Source: Bangladesh Bank Financial Sector Review, May 2006 8.3 Immediately after independence, the Government decided to nationalize the banking sector. The legacy of that move carried onfor several decades. Although two of the six NCBs were privatized in the 1980s and private banking was permitted, Bangladesh started out the 21'' century with a banlung system dominated by government-ownedbanks and several weak privately-owned banks. 8.4 The structure of the banking sector has evolved quite remarkably since 2001, benefitingfrom some bold reforms that have sought to deemphasize the role of the NCBs, strengthen competitive pressures, loosen government control, and tighten prudential regulations and regulatory quality (See 15'Financial Performance and Soundness Indicators of South Asia, World Bank, 2004. Annex 8.1 for a chronology of major banking sector reforms). The structural shift is most visible in the declining share of the NCBs in the industry's total assets, from 54 percent in 1998 to 37 percent in 2005, mirrored by the increase in the share of the PCBs, from 27 percent in 1998 to 46 percent in 2005. In 2004, for the first time, the combined share of the NCBs and DFIs in total banking sector assets was lower than the combined share of the PCBs and FCBs. Similar trends can be seen for bank deposits and loans; with the PCBs and FCBs increasing their market share at the expense of the NCBs and DFIs. 8.5 A number of key performance measures in the banking sector have responded positively to the recent reforms (Table 8.2). The Capital Adequacy Ratio (CAR) of banks increased from 6.7 percent in 2001 to 8.3 percent in 2006, mainly due to sharp increases in the CARSof FCBs and DFIs; the CAR of the NCBs fell over the period. The non-performing loans (NPLs) to total loans ratio improved considerably for all categories of banks, responding to restructuring measures, improved lending practices, and strengthened loan recovery drives. The gross NPL ratio of the NCBs fell from 46 percent in 1999 to 23 percent in 2005. Underlying the history of the high NPLs in the NCBs and DFIs are the problems of weak management, constant political interference, and problems of corruption and directed lending; issues which are beginning to be addressed under the ongoing reforms. The PCBs managed to cut their gross NPL ratio to 5 percent by 2006, down from an astounding 33 percent in 1998. There is substantial variation in performance among the PCBs though, with several older ones having been under intensive Bangladesh Bank care for over a decade due while others have maintained healthy balance sheets positions.'59 Table 8.2: PerformanceMeasure in the Banking Sector Capital to risk weighted assets ratio Ratio of gross NPL to total loans Bank Type 2000 2001 2002 2003 2004 2005 2006 2000 2001 2002 2003 2004 2005 2006 NCBS 4.4 4.2 4.1 4.3 4.1 -0.4* 1.1 38.6 37.0 33.7 29.0 25.3 21.3 22.9 DFIs 3.2 3.9 6.9 7.7 9.1 9.2 9.5 62.6 61.8 56.2 47.4 42.9 34.9 33.7 PCBs 10.9 9.9 9.7 10.5 10.3 9.2 9.8 22.0 17.0 16.4 12.4 8.5 5.6 5.4 FCBs 18.4 16.8 21.4 22.9 24.2 25.1 22.7 3.4 3.3 2.6 2.7 1.5 1.3 0.8 Total 6.7 6.7 7.5 8.4 8.7 7 3 8.3 34.9 31.5 28.0 22.1 17.6 13.5 13.1 Note: NPL numbers are without adjustment for actual provision and interest suspense. * Negative CAR for NCBs in 2005 because of negative capital of Agrani bank Source: Bangladesh Bank Quarterly, various issues. 8.6 Despite the recent improvements, the banking sector, having been plagued by systemic distress over the last 20 years, is only beginning to emergefrom a crisis situation and has a long road ahead to full recovery. Applying one of the criterion for a systemic banking crisis, gross NPL ratio of more than 10 percent of aggregate loan portfolio, Bangladesh has been in one since 1983 - with gross NPLs standing at 13.1 percent in 2006 and NPLs net of loan provisions at 7.1 percent of total 10ans.l~~ In addition, provisioning by the NCBs and the problem PCBs has typically fallen well short of requirement. The required to actual provisioning ratio for the banking sector as a whole fell from 60 percent in 2001 to 41 percent in 2004, before recovering a bit to 48 percent in 2005. With a provision maintenance ratio of about 25 percent and a provisioning shortfall equaling Tk.40 billion, the (in)solvency situation was the worst in NCBs. The FCBs were the only category that met its provisioning requirement in 2005. lS9There were several waves of PCBs licensed at different points in time, referred to as generations. The first generation of PCBs was licensed during 1972 and 1989, the second during 1991 and 1996 and the third between 1999and 2001. Other criteria commonly used to identify a systemic banking crisis include: (i) emergency measures were taken to assist the banking system (such as bank holidays, deposit freezes, blanket guarantees to depositors or other bank creditors), (ii) large-scale nationalizations took place, or (iii) the fiscal cost of the rescue operations was at least 2 percent of GDP. Compare Demirguc-Kunt and Detragiache (2005). In. Bangladesh's Financial System in International Comparison 8.7 The level of development of Bangladesh's banking system is comparable with other South Asian countries and higher than the average for low-income countries, although on some important indicators of banking efficiency and solvency Bangladesh lags many comparators. Other segments of the financial system, such as the insurance sector and the stock market are substantially less developed in Bangladesh than in the comparator countries. Both the banking system and stock markets are less developed and less efficient in Bangladesh than in East Asia. 8.8 As seen in Table 8.3, in 2004, on all three financial intermediation development indicators used - Private Credit to GDP ratio, Liquid Liabilities to GDP ratio, and Bank Deposits to GDP ratio - Bangladesh performed worse than India, but better than Indonesia, Pakistan, the average for South Asian countries, and the averages for low-income countries. Not surprisingly, it fared worse than the average for East Asian countries whose financial systems have by now recovered from the 1997 crisis. On a positive note, Bangladesh's loan-deposit ratio stood at an exceptionally high 80 percent, i.e. four-fifths of banks' deposits were intermediated into private sector loans. This is even higher than the East-Asian average. One of the reasons for this is that the Government directly competes with banks through the National Savings Schemes (NSS), which go straight into its budget, leaving limited need for direct bank loans to the ~overnment.'~' Table 83. Banking and Non-Banking Indicators for Bangladesh in International Comparison (2004) I I Panel A -Banking Sector Indicators Loan-Deposit Liquid Private Bank Overhead Net Ratio LiabilitiestGDP Credit! DepositsIGDP Costs Interest GDP Margin Bangladesh 0.801 0.389 0.274 0.342 0.026 0.026 Indonesia 0.540 0.433 0.210 0.389 0.031 0.048 India 0.641 0.616 0.328 0.51 1 0.022 0.032 Nepal 0.367 0.021 0.033 Pakistan 0.748 0.469 0.257 0.344 0.02 1 0.028 Low Income 0.656 0.294 0.147 0.220 0.058 0.077 South Asia 0.677 0.486 0.269 0.402 0.024 0.03 1 East Asia & Pacific 0.7 12 0.544 0.402 0.456 0.026 0.041 Panel B -Non-Banking Sector Indicators Stock Market Value Stock Market Life InsurancePremium Non-Life Insurance TradedlGDP TurnoverRatio Volume/GDP Premium Volume/GDP Bangladesh 0.016 0.365 0.004 0.002 Indonesia 0.107 0.430 0.006 0.007 India 0.548 1.137 0.024 0.006 Nepal Pakistan Low Income South Asia East Asia & Pacific 0.223 0.311 I 0.0 17 0.011 1 Source: Financial Struc we Database, Beck, Demirguc-Kunt and Levine (2000). Data are revised up to 2004. 16' These NSS are mostly for 3-5 years of maturity and their sale is restricted to individuals. Their rate of return is significantly above that of bank deposits, especially since there seems a very low penalty for early liquidation. At the end of 2004, there were Tk 349 billion outstanding NSS, but only Tk 182billion outstandinggovernmentpapers. 8.9 Comparisons of operational efficiencypaint a mixedpicture. The NPL ratios in Bangladesh are significantlyhigher than in the comparator countries. In 2002, the gross NPL ratio of the banlung sector stood at 28 percent, compared with 9 percent in India, 16percent in Pakistan, and 14percent in Sri Lanka. Despite the reductions since, Bangladesh has not yet caught up with these comparator countries. On the other hand, overhead costs and net interestmargins-measures of inefficiency in the banking system-are at comparable levels with other South Asian and East Asian countries and significantly lower than the average for low-incomecountries(Table 8.3).Ib2 8.10 Comparisons with other countries in the region and low-income countries suggests that banking penetration in Bangladesh is adequate for the level of financial and economic development. Here, we use a new database on access to and use of banking services, compiled from regulatory entities and public information for the years 2003/2004 and discussed in Beck, Demirguc-Kunt and Martinez Peria (2005). Bangladesh has 47 bank branches per 1,000 square kilometer which is significantlyhigher than in comparator countries (Figure 8.1). Bangladesh's demographic branch penetration, four branches per 100,000 people, is at similar or higher levels than in Pakistan, Nepal and the average low-income country,but significantly below the level observed for India, Indonesia and Sri Lanka. ATM penetration of less than one ATM per 1,000 square kilometer and per one million people is significantlylower than the comparator countries with the exception of Nepal. The number of loan and deposit accounts per capita is higher than in Pakistan and the average low-income country, while the average loan size and deposit account size relative to GDP per capita -negative indicatorsof outreach-are below the levels in Pakistan and the average low-income country. While these indicators are only for the formal banking sector, the limited data available for the microfinance sector suggest that access to and use of formal financial servicesis higher than in many comparablecountries. Bangladesh has the highest microfinance penetration in the world (borrowing clients constitute 13 percent of total population), twice that of Indonesia, which has the second highest penetrationrate.163 Figure 8.1: Financial SectorPenetration Branch Penetrationkross Countries A N PenetrationAcross Countries Branches per ASMSper 1.000 1,000 sq km sq km I IIndonesia IIndonesia I Lanka Sri ISriLanka mNepal I Nepal Number of IPakistan N u W r ofASMs branches per w mr 100.000 Pakistan Bangladesh 100,000 people ILowIncomeCountries I l l ILowIncomeCountries 162Overhead costs are banks' total operating costs relative to total assets while net interest margins are net interest revenue (i.e. interestrevenue from loans minus interest costs from deposits) relative to total earning assets. 163Honohan (2004). 94 1 f I LoansIDeposit per CapitaAcross Countries Ratio of LoanlDewsit Size to Income Ratio Across Countries I BPakistan t I @Bangladesh Number of i BLowincomeCountries Deposits per 1000People i 8 Number of l Pakistan Average Loan Loans Per 1000 Bangladesh SizelGDP Per People · LowIncomeCounlries CaP i I I 0 50 100 150 200 250 0 5 10 15 Source: Beck, Demirguc-Kunt and Martinez Peria (2005) 8.11 Other segments of thefinancial system in Bangladesh are significantly less developed than the banking system, and this disparity stands out relative to comparator countries. There is a small but growing non-bank financial sector comprising leasing and finance companies. The stock market is a small, inactive and stagnant market. Stock market capitalization relative to GDP stands at about 5 percent, and the turnover ratio - trading relative to capitalization - is lower than most comparator countries. The shares of financial institutions - mandated by law to list - dominate the Dhaka stock exchange;164in December 2006, they constituted 55 percent of market capitalization. There were only two Initial Public Offerings (IPOs) of non-financial companies in 2005 and none in 2004. There are few incentives for a firm to go public, and stocks of a typical company are rarely traded 8.12 Long-term contractual savings institutions are virtually absent and even their miniscule presence does not serve the societal purpose of promoting long-term saving. The insurance sector is small and inefficient - in 2004, life and non-life insurance penetration (premiums relative to GDP) stood at just 0.4 percent and 0.2 percent, respectively, below the average for both low-income and the South Asian country averages (Table 8.3). Not only that, life insurance companies undertake reverse maturity transformation, i.e. transform long-term liabilities into short-term assets, such as bank deposits, which can be partly explained by the reluctance to issue long-term savings instruments that would have to compete with the NSS. There are few private pension funds and mutual funds whose development is most likely also prevented by the existence of the NSS that offer rates above deposit interest rates in the banlung sector. Both the pension and mutual fund industry are dominated by public pension schemes and by the Investment Company of Bangladesh (ICB) that also holds a large part of the shares on the Dhaka Stock ~ x c h a n ~ e Unlike private funds, ICB is not subject to Securities and Exchange Commission (SEC) . ' ~ ~ supervision and has fewer restrictions on its activities, so that private institutions do not face a level playing field.Ib7 This lack of privately owned institutional investors has negative repercussions for corporate governance and stock market liquidity. 8.13 The picture painted by the aggregate indicators is confirmed by firm-level indicators of financing patterns. Firms in Bangladesh finance a lower share of their worlung capital and new 164There is a second smaller stock exchange in Chittagong. 165Sobhan and Werner, 2003. 166It is estimated that ICB holds 20% of shares through their mutual funds and own accounts as brokerage firm (AsianDevelopment Bank, 2004) 167Sobhanand Werner (2003). investment with retained earnings than firms in Pakistan and the average firm in low-income country, but more than in other comparator countries (Figures 8.2.a and Annex Figure 8.2). Differentiating by sources of external finance, Bangladeshi firms finance a larger share of working capital and new investment with banking credit, while private equity and the sales of stocks play a miniscule role. At the same time though, Bangladeshi firms view access to external finance and cost of external finance as bigger obstacles than f m s in Palustan, Sri Lanka, Indonesia and the average firm in the average South Asian or East Asian country or, for that matter, in the average low-income country (Figure 8.~.b).'~~ Figure 8.2a: Sources of Firm Financing: Share of Figure 8.2.bFinancing Obstacles: Access to Working Capital and Cost of Financing I Intermi fwds r .Equity ard sale of stock i0 =N3 Ohtact 1= Mim Obstacle;2 = ModwaleObstacle;' BTradeuedit (supptiwor custom credit) 4 = Mapr Obstacle;4 = Vgr Severe Obstacle 1 100% ; Slnforrralsources '0° Comrcial bank kans (bcal and foreign) .mess a i i m ~- e *.Gihersources BCostofFinance Source: Investment ClimateAssessments IV. Finance Development and Growth in Bangladesh: Is there a Link? 8.14 Financial development indicators, such as Private Credit to GLW, Total Deposits to GDP and M2 to GDP display rising long-term trends in Bangladesh, suggesting financial sector deepening (Figure 8.3 and Annex Table 8.2). As seen in Figure 8.3, the year 1991 marked a turn-around in financial sector performance, resulting from the Financial Sector Reform Program (FRSP) that was initiated in that year and aimed at moving the financial system to a more market-based one. Reductions in interest rate controls and credit quotas were among the major changes under the FRSP. It is important to note, however, that the FSRP led to a more quantitative improvement in the financial development indicators: i.e., although, for example, the volume of credit increased in the first phase, the quality of lending did not necessarily improve. As noted in Section 11, the qualitative shift occurred as a result of the reforms started in 2001. Although these are subjective assessments, that Beck, Demirguc-Kunt and Levine (2006) fmd that in countries where firms report higher financing obstacles, industries that depend more on external finance grow relatively more slowly and investment resources are reallocated more slowly as demand changes across industries. Also, Beck, Dernirguc-Kunt and Maksimovic (2005) fmd a significant negative impact of financing obstacles on firm growth. Figure 83: Financial Development,Investment and Income in Bangladesh over Time -o- cr-y -u- dep-y -A- m2-y -x- i-y +y-pcap c r y is Private Credit to GDP, depy is total deposits to GDP, 1x12-y is M2 to GDP, i y the investment-GDP ratio and yqcap income per capita in current USD. Source: Bangladesh Bank and authors' calculations. 8.15 Financial development has been positively correlated with and very likely has contributed to the rising income levels. Figure 8.3 shows the positive correlation between the financial development indicators and the investment to GDP ratio as well as GDP per capita. Use of econometric tim-series techniques - that are elaborated in Annex 8.2 - shows that not only is financial development positively correlated with investment and income in Bangladesh, it also seems to have positively impacted the two: i.e., higher levels of financial development have caused higher levels of investment rate and per-capita GDP. Results show that a one percent increase in the ratio of private credit to GDP led to a 0.15 percentage point increase in the investment rate and a 0.22 percent increase in per capita income. The finding of a positive impact of financial development on per capita income is after controlling for the indirect impact through investment activity, suggesting that financial development likely enhanced both the level and efficiencyof in~estment.'~~ 8.16 Thepositive relationship betweenfinancial development, investment, andper capita income is driven by the post-FSRP period of 1991 to 2005. The econometric results reported in the previous paragraph hold only for the 1991-2005period and not for the pre-FSRP period of 1976to 1991. This is consistent with the cross-country evidence discussed below and canies some important policy conclusions. Specifically, a financial system that relies heavily on nationalized banks, interest rate controls and directed credit, as was the case in pre-FSRP period, cannot be expected to fulfill its importantrole in channeling society's savings into investmentsand fosteringeconomicgrowth. V. Finance as Key to Faster Economic Development:Cross-Country Evidence 8.17 Thepositive impact offinancial development on income levels in Bangladesh is consistent with economic theory as well as cross-country evidence. Financial markets and institutions arise to alleviate market frictionsthat prevent the directpooling of society's savings and channelingto investmentprojects. Well developed financial systems ease the exchange of goods and services by providing payment services, help mobilize and pool savings from a large number of investors, acquire and process information about enterprises and possible investment projects, thus allocating society's savings to its 16'The results also show a positive and significant relationshipbetween investment and per capita income. In the long-run a 1percentage point increase in the investment-GDPratio is shownto increase per capita income by 0.2%. most productive use, monitor investments and exert corporate governance and help diversify and reduce liquidity and intertemporalrisk.170 8.18 A well-established body of evidence corroborates that countries with better develo ed financial systems experiencefaster economic Annex Table 8.1 and - Figure 8.4 summarize this evidence; countries with higher . S ...:*a. levels of credit to the private sector relative to GDP c,.' experienced hi" per capita growth rates over the period 1980 to 2003."~~"~mote that the positive relationship s. e a . . p- l "8 . ~:2*: between private credit and growth shown in Figure 8.4 is 8 · . after controlling for the standard growth regressors such as GF- ·.* .** education, government consumption, trade, black market premium, inflation, and population growth.] This Y, r- relationship is not only statistically, but also economically 3 -2 R 1~ O ~ J I I X ) o I 2 significant. Results in Annex Table 8.1 suggest that a doubling of private sector lending to GDP ratio in Bangladesh over the period 1980-2003,from 22 to 44 percent, would have resulted in 0.7 percentage higher GDP growth per year.'74.'75 Other measures of financial development, such as M2 to GDP or the turnover ratio in stock markets yield similar re~u1ts.l~~ 8.19 Financial development is not onlypro-growth, but also pro-poor. Annex Figure 8.1 shows the negative relationship between Private Credit and the average annual change in Headcount - the share of population living on less than one dollar a day -over the period 1980-2000.'~~Note that the relationship represented in Annex Figure 8.1 is after controlling for the effect of average economic growth suggesting that financial intermediarieshelp the poor also by improving the inequality ~ituati0n.l~~ 8.20 Bigger is not always better infinancial development, however! Forfinancial development to positively impact growth, an enabling policy environment is essential. Private credit is not a policy variable itself. A rapid increase in private sector lending not supported by macroeconomic stability and the contractual and informational framework discussed below can result in fragility and banking crises, as experienced by many countries over the past 30 plus years, with associated costs in the form of lost growth and worsened poverty situation.17' 170Levine, 1997and 2005. 17'TObe sure, there is also a feedback of faster economic growth on the development of the financialsystem. '72Private Credit to GDP is a standard measure of financial intermediarydevelopment and is the ratio of claims by deposit money banks and other financial institutions on the private, domestic non-financial sector relative to GDP. Private credit to GDP was 173%in the U.S. in 2003, but only 2% in Mozambique. 173The set of explanatory variables follows a standard regression set-up from the finance and growth literature, for example, as used in Beck, Levine and Loayza (2000). 174Such counterfactualshave to be interpreted with caution as the coefficients indicatemarginal effects from which we are extrapolating to discrete changes. Further, this counterexample serves an illustrative purpose; it does not actually tell us how to achievehigher growth rates. 17'The relationship between financial intermediary development and economic growth is not only robust to controlling for other factors that are associated with economic growth, but to controlling for the reverse causation from faster economic growth to financial developmentand to the relationshipbeing driven by a third factor. '76Levine, Loayza and Beck (2000); Beck and Levine (2004). '77AS shown by Beck et al. (2004), this relationship is robust to controlling for characteristics such as government consumption,inflation, openness and population growth, as well as for reverse causationand simultaneitybias. 17'Changes in poverty can be decomposed into economic growth and changes in income inequality (Bourguignon, 2004). Kraay (2006) shows that the larger part of poverty changes can be explained by economic growth. 17'Dernirguc-Kunt and Detragiache(1999); and Caprio and Klingebiel(1997). VI. Three Pillars of an Effective and Stable Financial 8.21 This section discusses the three building blocks - macroeconomic stability, contractual framework, and informational framework - for a sound and effective financial system and benchmarks Bangladesh in those areas. Bangladesh has made impressive progress in securing macroeconomic stability,but still has a long way to go in building the contractual and informational framework. Macroeconomicstability 8.22 Macroeconomic stability is a prerequisite for financial development and afirst building block for an effectivefinancial system. A low and stable rate of inflation provides incentives for financial rather than non-financial forms of savings. It is also conducive to long-term contracting and thus long- term savings and investment. Economic theory predicts and cross-country studies have shown that countries with lower and more stable inflation rates have more rapid banking sector and stock market development.181This is illustrated in Annex Figure 8.3, which plots Private Credit to GDP over the period 1980 to 2003 against the average annual inflation rate for the same period for a cross-section of countries. Similarly, a lower fiscal deficit can enhance financial sector deepening by avoiding crowding- out of private sector lending and investment. As discussed in Volume I of this report, macroeconomic stabilityhas been an area of strength for Bangladesh, with inflation rates having stayed in single digits for almost two decades now and the fiscal deficit and debt situation well under control. Contractualframework 8.23 Financial contracts depend on the certainty of legal rights of borrowers, creditors and outside investors and the predictability and speed of their fair and impartial enforcement. Private property rights and enforcement of contracts - both vis-A-vis other private parties and vis-A-vis the government - are thus a second crucial pillar of an effective financial system. Savers will relinquish control over their savings only if they can be sure to receive principal and contracted interest in return. Similarly, outside financiers- be they equity or debt holders - will only be willing to invest in companies and projects if their legal claims and rights are upheld. Critical in the relationship between minority shareholders and controlling shareholders and management are often problems of looting and tunneling, i.e. the possibility for the controlling shareholder to remove assets outside the firms, and the remedial actions that the legal system offers minority shareholders against them.''' This points to both creditor rights and minority shareholder rights as being critical for external financing, as well as the role of corporate governance in both financial institutionsand non-financial corporations. 8.24 An effective legal system is an importantfeature of the contractualframework as it provides timely, efficient and impartial resolution of insolvent borrowers through the efficient liquidation of unviable enterprises and the rehabilitation of viable businesses. International comparisons have provided ample evidence for the critical role of legal system efficiency and its different elements for financial sector deve10~rnent.l~~The effect of legal system efficiency on financial intermediation is illustrated in Annex Figures 8.4 to 8.6, which show the positive relationship between creditor rights and private credit to GDP, the negative relationship between the time it takes to enforce contractsand private credit to GDP and the positive relationship between better protection of minority shareholder rights and the market capitalization in stock markets. This and the followingsection draw heavily on Beck (2006). 181Boyd, Levine and Smith (2001). Johnson et al. (2000). La Porta et al. (1997); see overview in Beck and Levine (2005). 8.25 While Bangladesh has made some progress in the contractualframework over the past years, there is still significant roomfor improvement. Corruption is widely regarded a major problem in the country (Chapter 7). Although creditor and minority shareholder rights are relatively strong on the book, their enforcement is weak. Lenders and borrowers have to deal with an ineffective collateral system for movable and immovable assets. Land titling is hampered by poor and unreliable records, high fees, and fraud. The movable collateral regime is effectively available only for limited liability companies, as these are the only ones that can register charges with the Registry of Joint Stock ~ o m ~ a n i e sThis . ' ~ ~effectively excludes most of the small and micro enterprises from using such assets as collateral. These problems have increased reliance of lenders on personal guarantees from company directors. The bankruptcy legislation is rarely used in practice as a tool to resolve corporate distress, with less than half of the cases in the two bankruptcy courts actually being declared bankrupt at the end of the process.'85 This is partly due to the stigma attached to bankruptcy and possible criminal prosecution of bankrupt debtors.lg6While some progress has been made with the recent introduction of the money loan courts and alternate dispute resolution mechanisms that allow a faster and less bureaucratic enforcement of claims, it is still too early to say whether these reforms have significantly impacted the availability and cost of loans. Further, appeals to the high court still pose a major bottleneck to the quick resolution of claim resolution because of the delays involved.'87 8.26 The deficiencies in the contractualframework are reflected in cross-country benchmarking. The 2006 Doing Business Indicators rank Bangladesh 147" out of 153 in terms of days needed to register a property, 76thout of 155 on days needed to enforce a contract, and 101'' out of out of 144 on days needed to close a business. Informational framework 8.27 Financial markets and institutions arise due to information asymmetries that prevent direct interaction between multiple savers and investors. Accordingly, effective financial intermediation depends on tools to reduce these information asymmetries. The informational framework is thus a third crucial pillar for effective financial intermediation. International experience has shown that credit information sharing is important to reduce adverse selection problems and foster competition in the financial system. Credit registries have an important role to play as they can dramatically reduce the time and costs of obtaining such information from individual sources. They are especially important for SMEs as their creditworthiness is harder to evaluate and they have less visibility and transparency relative to large enterprises. International comparisons show that countries with better developed systems of credit information sharing enjoy higher levels of financial development and their firms report lower financing obstacle^.'^^ Annex Figure 8.7 illustrates the positive relationship between the efficiency of credit information sharing and Pnvate Credit to GDP. 8.28 Bangladesh has a well functioning public Credit Information Bureau (CIB), established in 1992 and housed in the Bangladesh Bank. It collects data on a monthly basis from all banks and financial institutions that are regulated and supervised by Bangladesh Bank on all borrowers. While there is no unique borrower identifier, the wealth of information of individual borrowers seems to allow a relative good match. All financial institutions are required by law to obtain information about a borrower from the CIB before granting or renewing a loan. This provision was introduced in the effort to reduce NPLs in the banking system. Meagher (1998). Ig5Asian DevelopmentBank, 2004. Ig6Sobhanand Werner (2003). Ig7ibid. 188Japelli and Pagano (2002); Love and Mylenko (2003);Djankov et al. (2005), and Miller (2003) for an overview. 8.29 While the CIB has a long and positive track record, several shortcomings of the existing financial informational system should be noted. First, the system is still limited by the lack of computerization. Second,banks effectively obtain current information only about existing borrowers,but not about past payment performance. This is an important omission in the CIB7sfunction as it prevents borrowers from building "reputation collateral", which would allow them to use their track record as reference in future loan applications. Third, only financial institutions participate in the information sharing agreement, other non-financial institutions such as utility companies, which are a source of valuable credit information about customers in other countries are excluded. Further, the minimum threshold of Tk 50,000 (US$700) excludes small and opaque enterprises. While technical problems are certainly recognized, a move towards includingall borrowers irrespective of the size of their loans should be envisioned in the medium-term. Another important step would be to link the existing CIB to the credit registry to be created for micro-finance institutions (MFIs). Finally, private institutions should be allowed to provide credit information services, for which the necessary legal provisions would need to be changed. While it seems unlikely that such an industry would immediately flourish, it can become an important element in the move from a relationship-based economy to an arms-length economy, which relies on numerous informationbrokers, includingcredit rating institutions and credit registries. 8.30 Similarly, more accurate and transparent financial statements can help reduce information asymmetries between borrowers and lenders. Financial statements that give an accurate picture of a firm's financial situation reduce screening and monitoring costs for financial institutionsand increase the efficiency of resource allocation. Cross-country comparisons have shown a positive association of more transparent and comprehensive accounting standards with higher levels of financial development.'89 Similarly,transparent and more informativefinancial statementslower information costs for investorsand can foster stockmarket development and liquidity.lgO 8.31 The quality of financial statements in Bangladesh, however, is undermined by weak accounting and auditing practices, due to institutional weaknesses in regulation, compliance, and enforcement of standards and rules.191While all incorporated companies must file annual audited financial statements with the Registrar of Joint StockCompanies, there is no effective enforcement of the timely and accurate filing. A review of published fmancial statements revealed numerous compliance gaps. There is limited accounting and auditing capacity and numerous violations of professional ethical standards. While listed companies are subject to additional oversight by the SEC, there are legal and judicial constraintsto effective enforcement of this oversight. Further, there seemslittle confidence in the financial reports issued by companies with no or perverse price effects of such anno~ncements.'~~. 193 More importantly, the costs of going public seemto surmounttheir benefits, as public f m s are subject to more scrutinyby tax authorities. VII. The Evolving Role of the Governmentin Bangladesh's Financial System 8.32 Whatis the appropriate rolefor the government in the development of thefinancial system, in addition to providing its three essential pillars described above? Should it leave it all to the markets, take it upon itself to provide financial services, or seek to enable markets by addressing market failures and reducing market frictions? As described in Box 8.1, it is evident from global experiences that the Is9Levine, Loayza and Beck (2000)and La Porta et al. (1997). 190Levine (2003). 19'Investment Climate Assessment, World Bank, 2003 '92Asian DevelopmentBank, 2004. '93Sobhanand Werner (2003)even quote the example where a bank's stock price increased after irregularitiesin its lendingpracticesbecamepublic, most likely due to manipulating transactionsin the market. (relatively new) market enabling approach is generally the most attractive option for government intervention. What has been the approach adopted in Bangladesh? Box 8.1: Three Different Approaches to the Role of Government in Financial Service Provision The role of the government in financial service provision has been controversial. While most economists agree by now on the role of government in providing a stable monetary environment and the contractual and informational frameworks, the government's roles as provider, regulator, and supervisor of financial service remains moot. Three differentapproachesto the government's role have been applied: The laissez-faire approach: This approach sees no role for government involvement in the financial system. Market discipline rules the relationship between banks and depositors and between banks and borrowers. A modified laissez-faire approach stresses the importance of macroeconomic stability and the contractual and informational framework and sees a role for government in providing these necessary elements. The laissez- faire approach has been criticized for ignoring market imperfections and failures that lead to a breakdown of bank-depositor and bank-borrowerrelationshipsas envisionedin theory. From market to government failures: Government solutions to overcome market failures rely on two crucial assumptions. First, governments know better than markets; second, governments act in the best interest of society. Both assumptions find little empirical support. Bureaucrats as bankers have failed almost everywhere, but especially in developing countries; economies with a higher share of government-owned banks typically have lower levels of financial development, more concentrated lending and lower economic growth and are more likely to suffer systemic banking sector fragility (La Porta et al, 2001; Barth, Caprio and Levine, 2004). High NPLs in these systems often result in banking crises (Barth, Caprio and Levine, 2004; Caprio and Martinez Peria, 2003), at large costs to the budget. Further, the negative impact of government-owned banks spreads oftenbeyond these institutionsand adverselyaffect governanceand efficiencyof private banks as well. Enabling markets -a new approach: According to this approach, rather than replacing market failures with government failures, policies need to address market failures directly by reducing market frictions. Cross- country comparisons show that countries with lower entry barriers into the financial system, fewer restrictions on banking and a higher share of foreign-owned banks, each resulting from lowering of market frictions by governments, have more competitive banking systems (Claessensand Laeven, 2004; Beck, Demirguc-Kunt and Maksimovic (2004), Demirguc-Kunt, Laeven and Levine (2004); Beck, Demirguc-Kunt and Levine (2006b). This is not a laissez-faire approach; rather, it relies on the government to enable and develop markets. The approach relies on private institutions to provide financial services efficiently but takes into account their incentive structure. It sees a critical role of the government in ensuring a competitive and contestable financial system The approach envisions monitoring and discipliningof banks by private actors, together with a role for effective government intervention in troubled institutions. It also argues for strengtheningthe independenceand accountability of regulators to protect them against political and regulatory capture and, a careful, incentive compatible designing of deposit insurance where it already exists. It sees an important role for the government in avoiding segmentation in the financial sector through expanding access to the payment system or the credit information sharing system beyond the commercial banks to bank-like institutions such as cooperatives or regulated microfinance institutions can help the financial system cater to marginal customers in all financial 8.33 The role of the government after independence was a hi hly interventionist one,following the examples of many other developed and emerging economie~.~~'Through nationalization of banks it hoped to enhance mobilization of savings, direct credit to "priority sectors", and make financial services affordable for larger parts of the population. Through interest rate controls it hoped to reduce lending costs, while credit quotas were supposed to guarantee flow of financial resources to priority and underserved sectors, such as agriculture. Through specialized institutions, such as agricultural banks and industrial development finance companies, the government hoped to provide more financial resources to these sectors. As in many other countries, deposit insurance was introduced to prevent bank runs and 194See Fry (1988) for a discussionof these policies and numerous examples in other countries. protect small savers. Essentially, in doing all these, the government completely rejected the laissez-faire approach and sought to replace it with a model based on government discipline. 8.34 Things didn't quite turn out as intended - marketfailures were soon replaced by government failures. This is also similar to the experience of other countries that attempted the public control approach: bureaucrats as bankers havefailed almost universally, but especially in developing countries (Box 8.1). By 1990s, it was clear that the highly interventionist approach had left a profound mark in the form of an inefficient state-dominatedbanking sector which fostered a culture of non-payment, provided excessive rents to private banks and distorted resource allocation at all levels; controlled interest rates that had nothing to do with risks and returns, resulting in gross misallocation of resources and large losses for the banking sector, and; political interference in most aspects of banking operations and supervision. By the late 1990s,the situation had come to such a pass that a 1998World Bank report noted that "The three institutional pillars of banlung - a strong regulatory system, well managed banks, and an effective court system - have crumbled to such a point that the banking institutions cannot be relied upon ensure the safety of deposits and efficiently allocate credit, their two most important functions. In a more difficult environment, the banking system could become Bangladesh's Achilles Heel." 8.35 Starting in 2001, the government began pulling back the banking sector from this brink by moving it toward a more market enabling approach. As described in Box 8.1, this approach recognizes the existence of frictions in financial markets which prevent their full functioning and calls for government intervention to create the conditions for a competitive and contestable system. Initial actions on reducing controls on the interest rates and credit quotas were undertaken under the FSRP in 1991,19' but it wasn't until 2001 when a more substantial reform program was begun leading to qualitative improvements in the banking sector performance, as reported in Section 11. 8.36 Authorities have embarked on a wide-ranging restructuring and divestiture program for the NCBs. Loan growth restrictions have been imposed on the four NCBs. Divestment of Rupali Bank, an NCB, to a strategic buyer is now close to being finalized. The three other NCBs are in the process of being corporatized and two of these would be partially privatized in the coming years; in the fourth NCB, the Sonali Bank, the government intends to sell only a minority share. The restructuring and divesture process, however, has been much slower than envisaged because of political interference, resistance from labor unions, and judicial roadblocks. Nonetheless, the expected successful divestment of Rupali Bank, in itself an important milestone for banking sector reforms, should provide impetus for and set a useful precedence in followingthrough with the rest of the NCBs. 8.37 Further, since 2001, Bangladesh Bank has followed international "best-practice" recommendations to strengthen bank regulations and supervision. Key measures include (see Annex 8.1 for a more complete set of banking reforms): The capital adequacy ratio was raised from 8% to 9% of risk-weighted assets and the minimum capital requirements from Tk 400 million (US$6 million) to Tk 1billion (US$15million). Loan classification, rescheduling, and provisioning rules were tightened and single exposure and large loan limits introduced. Risk management guidelineswere issued. 195Starting with the FSRP in the early 1990s, the government abolished most interest rate controls, with the exception of loans supporting export activities. Banks were asked to report reference lending interest rates for different sectors. Individual lending rates could only move within a band of 1.5 percentage points above and below this reference rate. Further, credit directives for privately owned banks were withdrawn, although government- owned banks were still expected to lend to priority sectors. Measures to improve corporate govemance were introduced. The number of board members in a financial institution was restricted to a maximum of 13;the term of a director was restricted to six years; shareholding by a single shareholder was limited to 10 percent of capital and only one member from a shareholding family could sit on the board; banks had to establish audit committees; limitations on divided payout were imposed as well as on expenditures for automobiles and travel expenses for executives. Responding robustly to the ensuing conducive environment, private banks have dramatically increased their market share and turned around their operational and financialperformance, as noted in Section11. 8.38 Certain amount of political capture of Bangladesh Bank, however, continues to vitiate market discipline and works against its move toward the market enabling approach. Bangladesh Bank has not been immune from political interference in the bank licensing process, often having to follow "recommendations" of the political class to allow new banks into the system. Fortunately, the practice has ceased in recent years. Nonetheless, the past practice has resulted in a number of weak PCBs, plagued by insider lending and other owner abuses. Political capture of the regulatory entity also prevents proper resolution of failing banks. 8.39 Although there is an explicit deposit insurance scheme, it has not been used. Rather, Bangladesh Bank has extended an implicit guarantee to all banks. Over the past years, no domestic bank has been allowed tofail;'96weak banks are referred to the Problem Bank Monitoring Department within Bangladesh Bank where they are subject to special supervisory oversight and certain regulatory restrictions and enjoy regulatory f0rbearan~e.I~~ 8.40 On its part, the Bangladesh Bank has sought to compensatefor the lack of market discipline (induced by the implicitguarantee) by taking a somewhat interventionist approach to supervision. For example, in order to expand outreach, it requires banks to open three banks in rural areas for every branch opened in an urban area. It has also issued job and task descriptions for board members and CEOs and made the appointment of a CEO subject to its approval. However, the resulting heavy-handed regulation has its own costs. For example, banks can get around the branching requirement by opening new branches just outside the urban areas - given that the urban areas in Bangladesh are still growing, soon these branches will be inside the urban area. This particular regulation, therefore, only causes sub- optimal allocation of bank branches. Regulations such as for branch opening or prescribing bonuses and benefits for board members and executives cause additional transaction costs and cause inefficiencies in banks' operations. Vetting the appointments of new CEOs by Bangladesh Banks seems rather in the old tradition of operating a financial system than regulating it. Similar govemance measures can, in fact, benefit the NCBs and were even expanded to cover them, but there is no enforcement for the NCBs. 8.41 Figure 8.5 illustrates the interaction of the government guarantee and supervisory approaches. On the horizontal axis is the degree of government guarantee and on the vertical axis the degree of supervisory intervention. In the past, absence of both market and supervisory discipline resulted in easy abuse by banks and high risk of fragility, placing Bangladesh in the South East comer. Since 2001, with the reform measures noted above, Bangladesh Bank has thus opted for the supervisory intervention approach (North East comer). Given the continuing explicit and implicit guarantee, politicization of licensing and the absence of exit of weak banks, this approach can avoid any "open" failures, but comes at huge efficiency cost - such as large interest rate spreads and sub-optimal allocation of resources -and involves high and increasing contingent losses for the government. Further, bank owners face incentives '96The only bank to be closed was BCCI. The bank was restructured and reopened as Eastern Bank. '97Pubali Bank spent 10 and National Bank 3 years under the intensified supervision of the Problem Bank MonitoringDepartment. to take aggressive credit risks, with little incentive for depositors or other creditors to exercise any market discipline in the presence of the implicit guarantee. This can explain the fragility in the banlung system even beyond government-ownedbanks in the 1990s. Figure 8.5: The Interactionof GovernmentGuarantees and SupervisoryApproach Easy abuse by banks Risk of fragility rises Governmentguarantee 8.42 Nonetheless, Bangladesh Bank has laid thefirst fundaments for market-based supervision. It requires banks to go public at the stock exchange and aim to list 50 percent of their shares. Also, disclosurerequirements are higher for financial institutions than for non-financial corporations:banks and non-bank financial institutions have to publish their annual financial statements in newspapers and display them for public view in their branches. Similarly, auditing requirements for banks are more stringent than for non-financial corporations; Bangladesh Bank maintains a separate list of approved auditors for banks. It is just that given its implicit guarantee for owners, creditors and depositors, these importantactions cannot serve their function in enhancingmarket discipline. VIII. A Financial SectorReform Agenda For Bangladesh 8.43 While some of the reform measures of the past few years are consistent with the market-enabling approach, others are not. Onthe whole, a general paradigm shift is needed. 8.44 The ongoing move to aprivately-ownedfinancial system is an important one. But as discussed in Box 8.2, international experience advises caution (but not necessarily delay): a failed privatization process can be worse than government ownership. Privatization is also not a panacea, as privatizing into a non-competitive environment where other elements of the framework discussed earlier are not in place would not bring the intended benefits. Key points to note to ensure success of the privatization process: First, such a process should be ideally undertaken by an entity that is independent from the political sphere and will undertake the choice of strategic investors purely on terms of merit not on socio-politicalterms. The entity that comes closest to this is the Bangladesh Bank. Second, the privatization process should be used to minimize the share of government ownership and if possible reduce it to zero. The future private owners might not object to or may even welcome minority government ownership. However, this would send the wrong signal, that the government would like to continue influencing decisions in the banks, and -even worse - stands ready to bail out the private buyers of the NCBs. While the first concern can be addressed by excluding government representatives fromNCB Boards, the second one is less easily to address. Third, the bank privatization process should be coupled with a resolution of the bad debts of the state-owned enterprises (SOEs). Finally, the government should ensure a level playing field between the privatized NCBs and other providers of financial s e ~ i c e s . ' ~ ~ Box 8.2: Bank PrivatizationProcess: Lessons from International Experience A failed privatization process can be worse than government ownership. Poorly designed and executed privatizationprocesses can lead to fragility and banking crises. Studies of privatizationprocesses have shown the benefits and pitfalls of privatizing banks (compare Clarke, Cull and Shirley 2005 and other papers on the same issue). Global experience shows that it is important to open the books of the entity being privatized early in the process. This would make the cost of government ownershipknown to the taxpaying public and generate political pressure for privatization. While privatization should not be delayed to minimize fiscal and efficiency losses, fire salesmight lead to divestment to investors that are not fit-and-proper or have not adequatelyassessed the situation of the bank. There are advantages to divesting to a strategic investor who can bring the necessary management and financial know-how, especially in a weak institutional environment and to open the privatization process to both qualified domestic and foreign bidders. By selling to reputable foreign banks, financial systems can benefit from expertise and from additionalresources. Finally, country studieshave shown the advantagesof full rather than partial privatizationto reduce possible future government influence. By giving up any ownershipstake, a government signalsthat it does not plan to intervene after privatizationand that it will . not bail out the new private investorsshouldthings go wrong. 8.45 Moving forward, de-politicization of entry into the banking sector and especially reform of the exitprocess would be key. The licensing process of the financial system has to be put on objective, non- political basis. While technical issues might impede proper bank failure resolution, such as the lack of a merger and acquisition provision in the Companies Act, it is especially the political willingness to resolve failing banks with market-based solutions that counts. Resolving a weak bank through liquidation or a purchase and assumption operation with the use of the explicit deposit insurance scheme would send an important signal to the market that Bangladesh Bank is moving away from an implicit guarantee for all financial institutions to a system where it cares about financial intermediation not about individual institutions. This does not imply embracing the idea of deposit insurance; rather it encourages the use of an existing but unused mechanism to move to a market-based bank failure resolution. 8.46 These different reform steps would help enhance market discipline and allow Bangladesh Bank to rely on a more market-based supervision. This can be illustrated in Figure 8.5 where divestiture of NCBs, de-politicization of licensing and a market-based bank failure resolution system would move the system away from the complete explicit and implicit guarantee to a system with a limited explicit depositor guarantee (North West comer). In the medium- to long-term, this would then allow Bangladesh to move towards the South West comer of the graph, a financial system with limited explicit guarantee and market-based supervision. Note that this would not mean a weaker Bangladesh Bank, but rather a regulator with the necessary powers and the necessary independence to foster market discipline and enhance a market-based financial system. 8.47 While the reform agenda laid out above implies a rather radical change in the approach to the relationship between supervisors and those supervised and between the government and the private sector, it also implies concrete initial steps that should accompany the currently ongoing divestiture process of the NCBs. First, gven the large number of banks in the market a temporary and limited moratorium on new bank licenses should be imposed. Second, during this moratorium, all banks should be subjected to special audits to assess their long-term viability. Third, banks that are found non-viable should be resolved, preferably through market-based solutions such as merger and acquisition or purchase and assumption operations. Fourth, the legal and regulatory framework for bank failure resolution should Ig8 This refers for example to Sonali Bank that has the function of a government agent in many localities, which effectivelygives it a dominant market position in these areas. be reviewed and possibly reformed. Finally, a medium- to long-term financial sector strategy should be developed that lays out further reforms in a clear and transparent manner. 8.48 Whilethe analysis andpolicy discussion and recommendations of this chapter havefocused on the banking system, they carry over to other parts of the financial system. As in the banking sector, regulation and supervisionin the other parts should also rely on market participants. Prudential regulation should be limited to deposit-taking institutions and be clearly separated from non-prudential regulation. Finally, a level playing field should be generated, which implies that financial institutions should be regulated and supervisedaccording to the products they offer, not according to the legal form they take. 8.49 Case in oint is the stock market, whichfaces supervisory problems similar to those of the banking sector." While the SEC tries to enhance governance by forcing companies to hold shareholder meetings and publish audited financial statements within certain timeframe, it also interferes directly in their decisions, by prescribing dividend levels, for example, under the assumption that low dividends even if justified by poor performance must be because of fraudulent management actions. The authority to ovemde the Companies Act or Articles of Association of any corporation is, in fact, provided under the law.200SEC can also interfere in the appointment of board members and CEOs, just as Bangladesh Bank can for banks. While the justification for this is the underdeveloped shareholder culture, cross-country comparisons show little evidence of a positive effect of public enforcement on stock market development, but a positive effect of private monitoring and disciplining: that is, laws that mandate disclosure and facilitate private enforcement through liability rules foster stock market development.201 8.50 Another causefor concern is the lack of a "limited liability culture". While Bangladeshi law recognizes limited liability companies, directors and owners who hold more than a certain share of equity are personally liable for losses beyond paid-in capital. The reliance of banks on personal guarantees of company directors due to the deficiencies in the collateral regime is a factor that undermines the limited liability culture. It is a huge disincentive then to serve as independent director on a company's board and a bamer to entrepreneurship. While this has to be understood in the context of a deficient contractual and informational framework with weak accounting standards and little protection for minority shareholders and creditors, it is clearly second-best and provides disincentives for innovation and entrepreneurship. IX. Conclusions 8.51 The financial sector is critical for sustained economic development and poverty reduction. This chapter has argued that beyond macroeconomic stability and an effective and reliable contractual and informational framework, the role of government has to be redefined to make Bangladesh's financial system more efficient and growth enhancing. Specifically, the government needs to move from the role of an operator and arbiter in the financial system to the role of an enabler and facilitator of markets. This would also facilitate the economy's move from being relationship-basedto an arms-length one. 8.52 Financial sector reforms, in fact, deserve a center stage place in the overall governance reforms agenda in Bangladesh, as elsewhere. "That is where the money is," as Willie Sutton, the famous bank robber, is purported to have said when asked why he robbed banks, and that is also where the temptation for corruption lies. A proper and transparent divestiture process of the NCBs and de-politicization of financial sector regulation and supervision can serve as a model in the divesture of other public sector entities. An autonomous and accountable Bangladesh Bank can similarly be a paragon for broader institutional reform. 199Asian DevelopmentBank, 2004; Sobhan and Werner (2003). 200Sobhanand Werner, 2003. 201La Porta, Lopez-de-Silanes and Shleifer,2006. Chapter 9: Infrastructure and Growth: a Focus on Power and Trade Facilitation 9.1 Infrastructure bottlenecks related to power and trade facilitation (ports and transportation) are severe in Bangladeshand place a major (and growing) drag on its economic performance. Only about 35 percent of the population - four-fifths of urban population and one-fifth of rural population - has access to network power, and load shedding to those fortunate enough to have access is rapidly increasing. Firm-level surveys consistently reveal access to power as being among the top obstacles to investment. Trade facilitation, critical for export competitiveness, is also not faring well. For one, Chittagongport, which handles nearly 85 percent of the country's merchandise trade, is plagued by labor problems, poor management, and lack of equipment. Moreover, transport networks connecting the ports with hinterland are underdeveloped and beset with logistical problems that impose significant costs on exporting and importing firms. This chapter presents an overview of these two infrastructure areas critical to sustained good growth in Bangladesh,and outlines a practical reform strategy for each. I. Power Sector:Weak Governance, Poor Outcomes 1.1. Introduction 9.2 Political economy factors have caused power to be chronically under-priced in Bangladesh, contributing to supply shortfalls for well over a decade. Unreliable power supply forces Bangladeshi industry to operate back-up generation, which, in most cases, is run on natural gas that is also heavily under-priced. Even then, it is estimated that firms pay up to 50 percent more for power they themselves generate compared to buying it off the public system grid, which undercuts their global competitiveness. Switchingbetween public grid and own generationdue to supply cuts is also not seamless-disruptionsin power supply result in significantproductivity losses in the manufacturingsector, as shown in Chapter 1. Moreover, this coping strategy may have also contributed to the concentration of industry in the Dhaka area, where gas is more readily availablethan, say, in the port city of Chittagong. 9.3 The defacto policy of using the country's limited gas supplies to offsetpower shortages, unless corrected, is likely to prove fatally flawed. Bangladesh is running out of natural gas, for commercial rather than geologic reasons; projections show that without significant power and gas sector reforms the countrycan experiencemajor gas shortagesas early as the next decade. 9.4 Recent government-commissioned studies indicate that new investment of US$1.5 billion may be needed annually to develop the gas andpower sectors. Clearly, the public sector is not in a position to undertake investment of this magnitude. Attracting large-scale private investment in gas exploration and production and in gas-fired power generationwill be key. To achieve that, it would be importantto: raise power tariffs -although in phases to give consumers time to adjust - and encourage further efficiency gains to enhancethe sector's self-financingcapacity; urgently address corruptionproblems plaguingthe power sector and tighten its governance; fill any remaining investment gaps (not covered by private investment) with donor financing and/or budget support; and take steps to manage demand growth better,primarily through more efficient pricing. Although not particularly controversial, these measures appear to be somewhat at odds with the political economy realities in Bangladesh. Nonetheless, the economic toll from failure to reform the energy sector is likely to be steep and can be considered among the main risks to the achievement of the 7 percent plus GDP growth target set by the Government in its Poverty Reduction Strategy (PRS).~'~ 1.2. Overview of Power Sector Trends and Issues I 1 9.5 Electricity consumption and generation remain quite low in Bangladesh, despite rapid growth since independence. Bangladesh's per-capita electricity consumption of 160kwh in 2005 (Figure 9.1) was among the lowest in the world. This is despite the fact that electricity generation has grown twenty-fold since independence (from under 1,000 GWh per annum to over 20,000 GWh), compared with the three-and-a-half fold increase in the country's real GDP. Installed capacity has grown to about 5,350 MW, of which 1,260MW is privately owned and operated. Largely because of unreliable state-owned power plants, it has been difficult to generate more than 3,700 MW at any gven time (Figure 9.2). Growth in both industrial and residential electricity sales has been strong, driven in large part by the success of the country's RMG sector and also by the rapid expansion of the Rural Electricity Boards's (REB's) household customer base. The result is a customer base that is dominated by these two categories, accounting for about 85 percent of total sales (Figure 9.3). Figure 9.2: Installed Capacityand Available Supply Figure 9.3: Electricity SalesTrends by CustomerType 4.000 3.500 14 3,000 2,500 2,003 1,500 6 1,000 4 5 M 2 0 0 . . . 8 W RecordedPeakW mEslimakd Shedding - loResidenSal rn hdustrial 0Colmercial 0Agriculture I Omer &=me: GPCB,l