TRANSPORT GLOBAL PRACTICE JUNE 2021 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING © 2021 International Bank for Reconstruction and Development/ The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Exec- utive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attri- bution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202- 522- 2625; e-mail: pubrights@worldbank.org. Cover image: The World Bank Group Contents Acknowledgments ................................................................................................................................................... 5 Executive Summary.................................................................................................................................................. 6 1. Introduction...........................................................................................................................................................13 2.  Greater Market Accessibility but Stagnant Rural Access..............................................................18 Existing Road Network ................................................................................................................................ 19 Market Accessibility...................................................................................................................................... 24 Rural Accessibility......................................................................................................................................... 27 Section Summary......................................................................................................................................... 32 3.  Wider Socioeconomic Benefits from Transport Connectivity .....................................................33 Greater Access to Domestic Market Increases Agriculture Production........................................................... 34 Better Access to Ports Augments Firm Productivity and Agglomeration Economies....................................... 38 Supporting Human Capital Development..................................................................................................... 44 Building More Resilience in Road Network.................................................................................................. 52 Section Summary......................................................................................................................................... 54 4.  Current Road Spending .................................................................................................................................55 Institutional Set-Up for Public Expenditure on Roads................................................................................... 56 How Much Is Needed? How Much Is Available?........................................................................................... 57 Where Does the Need Exist? Where Does the Government Spend?.............................................................. 59 What Is the Government Spending?............................................................................................................. 62 Section Summary......................................................................................................................................... 64 5.  Toward More Sustainable Road Financing ............................................................................................65 Further Strategic Prioritization...................................................................................................................... 66 Increasing Efficiency in Budget Execution..................................................................................................... 69 Increasing Efficiency in Road Procurement . ................................................................................................ 70 Optimizing Expenditures Given Limited Resources....................................................................................... 72 Investing in Climate Resilience..................................................................................................................... 75 Increasing and Diversifying Road Sector Revenues...................................................................................... 78 Section Summary......................................................................................................................................... 85 6. Annexes.................................................................................................................................................................86 ANNEX A. LITERATURE REVIEW ......................................................................................................................................87 APPENDIX B. SET-UP FOR ROAD NETWORK EVALUATION TOOLS ANALYSIS .................................................91 Road Network Data...................................................................................................................................... 91 Unit Road Preservation Project Costs............................................................................................................ 98 Vehicle Fleet Characteristics.......................................................................................................................... 99 References...............................................................................................................................................................100 4 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Image by John Hogg/The World Bank Group 5 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Acknowledgments The World Bank team led by Atsushi Iimi (Senior Economist, Transport Global Knowledge and Expertise Unit) and composed of Rodrigo Archondo-Callao (Senior Highway Engineer), Julien Morel (Transport Consultant), Rakesh Tripathi (Senior Transport Specialist), Nargis Ryskulova (Senior Transport Specialist), Pedro Chilengue (Transport consultant), and Desta Wolde Woldeargey (Program Assistant) prepared this study. Idah Pswarayi-Riddihough (Country Director), Maria Marcela Silva (Practice Manager), and Raymond Bourdeaux (Operations Manager) pro- vided overall guidance. The team acknowledges valuable contributions, advice, and technical inputs from the following World Bank staff: Diego Arias Carballo (Lead Agriculture Economist), James Markland (Senior Transport Specialist), Fiseha Haile (Senior Economist), Kulwinder Rao (Lead Transport Specialist), Xavier Espinet Alegre (Transport Economist), Car- olin Geginat (Lead Economist), Albert Pijuan Sala (Senior Economist), Shireen Mahdi (Senior Economist), Carlos Da Maia (Economist), Javier Baez (Senior Economist), and Zayra Romo (Lead Energy Specialist). The team is also grateful to Vivien Foster (Chief Economist, Infrastructure Practice Group) and InfraXchange seminar participants for their insightful comments and suggestions. Finally, the team would like to express special thanks to the government of Mozambique, including the Ministry of Public Works, Housing, and Water Resources (Ministérió de Obras Públicas, Habitação e Recursos Hídricos), the Road Agency (Administração Nacional de Estradas) and the Road Fund (Fundo do Estradas), for their close collaboration in data collection and analysis. 6 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Executive Summary 7 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING BACKGROUND 1. Until recently, Mozambique has achieved robust Figure E.1. Contribution to Economic Growth economic growth, mainly driven by the extractive and of Various Sectors service industries. The country has economic potential 18 Service secgtor in a wide range of sectors, such as agriculture, agro- 16 Non manufacturing industry Manufacturing business, light manufacturing, and mining. In 2018, it 14 Agriculture Growth contribution (%) also started producing liquefied natural gas along the 12 10 northern coastline, but since the debt crisis in 2016, 8 Mozambique has faced renewed challenges to sustain- 6 able, inclusive growth (figure E.1). The crisis revealed 4 the country’s weak governance in the public sector and 2 the importance of addressing sustainable, inclusive 0 1999 2000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 growth; job creation; local business development; and 2001 regional inequality in the country. 2. In Mozambique, poverty was reduced in 2014/15, although it remains persistently high in rural areas. Approximately 12.3 million people live below the pov- Figure E.2. Distance to Amenities: Rural vs. Urban erty line. In recent years, regional inequalities have widened. This is partly attributable to misallocation 100 2009 2015 90 of public infrastructure investment between urban Walking time (minutes) 80 and rural areas (World Bank 2019a). Rural popula- 70 60 tions in the northern and inland provinces are espe- 50 cially behind in transport connectivity. Many farmers 40 30 are still largely subsistence farming, lacking access to 20 market opportunities (figure E.2). Many companies in 10 0 the extractive industry are foreign owned, which pre- Rural Urban Rural Urban Rural Urban Rural Urban vents a trickle-down effect on the domestic economy. Transport Market Primary school Clinic As a result, the economy cannot generate sufficient good-paying jobs, especially for urban youth, creating a very thick informal sector. 3. Recent external shocks have added to the eco- to the global economy. More than 60,000 cases have nomic challenges. Mozambique is highly vulnerable been reported in Mozambique, and more than 700 to natural disasters. Cyclones Idai and Kenneth hit people have died.1 There is significant uncertainty the country consecutively in 2019, causing devastat- about the path of economic recovery from the pan- ing damage to the economy of approximately USD1.5 demic. Inclusive growth with local business develop- billion (10 percent of gross domestic product (GDP)). ment is called for more than ever. The COVID-19 crisis is bringing significant uncertainty 1 As of March 10, 2021, according to https://mz.usembassy.gov/covid-19-information/. Existing roads (km) 8 Financial needs ($ mil) INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Rehab/ Reconst needs (km) 0% 20% 40% 60% 80% 100% Zambezia Nampula Niassa Tete Inhambane Cabo Delgado Gaza Manica Sofala Maputo 4. Efficient, reliable transport connectivity is among Figure E.3. Road Sector Spending the most important conditions necessary for sustain- 800 4.0 (US$ million) able, inclusive growth. Greater transport connectivity 700 3.5 Road sector spending (US$ % of GDP can lower transport costs and time, facilitating efficient 600 3.0 Percent of GDP (%) 500 2.5 mobility, creating jobs, and thereby alleviating poverty. million) 400 2.0 It also increases access to facilities such as schools 300 1.5 and hospitals, which increases human capital devel- 200 1.0 opment. The road network is lifeline infrastructure. 100 0.5 More resilience must be built into the transport sys- 0 0.0 2010 2011 2012 2013 2014 2015 2016 2017 2018 tem. Cyclones Idai and Kenneth damaged 3,600 km of roads. The poorly maintained road network increases the vulnerability of the economy, amplifying such dam- 8. Market access and rural accessibility. Recent age and delaying postdisaster recovery. intensive investments in primary roads have substan- tially increased domestic market accessibility. Although 5. Objectives. Given this background, the objective of market access increased, in particular for people this report is to examine how the country can ensure living around large cities along major highways, the sustainability of investment in and maintenance such as National Highway N1, some regions, such as of roads while maximizing expected socioeconomic Niassa, Tete, and Gaza, have been left behind (figure benefits from road network development. The report E.4). Rural connectivity has long been a challenge in updates the status of the road network and recent Mozambique. The rural access index deteriorated from developments in connectivity, examines major socio- 27 percent in 2006 to 20.4 percent in 2010 and 19.3 economic benefits from greater connectivity, reviews percent in 2017 (figure E.5). It is estimated that approx- how the government has been spending its available imately 16 million people have no access to the road resources, and explores potential ways to increase the network in the country. efficiency and sustainability of road sector spending. 9. Wider socioeconomic benefits. Mozambique has 6. Current road network. Road transport is the most benefited from greater transport connectivity, but the important mode of transport in Mozambique, carry- effects vary from sector to sector. In the agricultural ing approximately half of freight traffic and 98 percent sector, average crop yields are higher where transport of passenger traffic. The current network covers the connectivity is better. Accessibility to domestic mar- whole country but is narrowly defined, with a low road kets is particularly important in Mozambique. In the density (3.8 km per 100 km2 of land). Before the debt industry and service sectors, local market accessibility crisis, the government invested heavily in the road net- is important as expected, but accessibility to ports is work, particularly in primary national roads (~USD400 also critical, suggesting the importance of regional and million per year; 2.5 percent of GDP) (figure E.3). global market access in stimulating productivity. 7. Primary paved roads are relatively well maintained, 10. Greater transport accessibility can induce people although some national roads are in poor condition, to visit health facilities when necessary, preventing particularly where climate vulnerability is high, for serious diseases such as malaria. This is an important instance, Cabo Delgado, Nampula, Niassa, and Zam- constraint, but other challenges must be addressed bezia provinces. Rural (unpaved) roads are generally simultaneously, such as affordability and sanitary and poor, with 44 percent in poor condition in 2017, a dwelling conditions. Various policies must be inte- decrease from 42 percent in 2010. grated to achieve development outcomes. 9 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure E.4. Transport Connectivity Figure E.5. Rural Access Index 11. Potential investment needs. Substantial financial with a peak of 64 percent of the total spending in the resources are needed to close the remaining gap in road sector in 2014. Mozambique has already followed transport connectivity: approximately USD940 million a third-generation road fund model, consolidating to repair and rehabilitate the existing road network and internal and external resources under the road fund. another USD10 billion to reclassify currently unclassi- Even if external resources (~60 percent) are excluded, fied roads to achieve universal access in rural areas. the country spent approximately USD200 million, which is still substantial. 12. Current spending. Mozambique has been dedicat- ing substantial fiscal resources to transport infrastruc- 13. After the debt crisis, available resources dimin- ture development—more than other countries in the ished, from about USD400 million (2 percent of GDP) in region. The government allocated on average more 2011-15 to USD200 million (1 percent of GDP) in 2018. than USD500 million (3.6 percent of GDP) annually to The government has prioritized road and bridge main- the road sector from 2010 to 2018. Before the debt cri- tenance, which is a good practice for preserving exist- sis, the government had spent a significant amount of ing road assets, but the need for road maintenance money on rehabilitating and upgrading national roads, continues to increase (figure E.6). 10 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure E.6. Percentage of Spending on Roads 15. More maintenance. Under-maintenance of the According to Activities road network is an increasing concern because of the 100% trend of decreasing internal resources and increasing 39.4 42.0 54.5 59.2 22.2 31.9 need for maintenance to preserve previous intensive 64.4 33.4 50% 28.9 capital investment. USD180 million would be needed 39.9 34.4 32.7 47.3 to preserve current road assets optimally, of which 21.8 27.8 23.8 28.0 18.0 0% approximately 70 percent should be allocated to peri- 2010 2011 2012 2013 2014 2015 2016 2017 2018 odic and routine maintenance. With current limited Administration Capacity building resources, the value of existing road assets will decline Maintenance of roads/bridges Construct/Rehab of bridges by 20 percent to 40 percent. In addition, resilient infra- Rehab/Upgrade of regional roads Rehab/Upgrade of national roads Project preparation Road safety structure is more costly, possibly increasing costs by Other projects PPP 10 percent to 100 percent. Approximately 11 percent of roads are in flood-prone areas. 14. Currently available resources are insufficient to 16. Road safety. Road safety has long been neglected maintain the road network properly. Approximately in Mozambique. Although mortality from road crashes USD182 million would be needed per year to achieve is high (32/100,000 people), government spending on maximum value from the current paved road net- road safety declined from approximately 2 percent of work and approximately USD400 million to maintain spending on building and maintaining roads in the the overall road network. With current resources early 2010s to 0.4 percent in 2018, far below the inter- (~USD200 million), it is likely that the road asset value national norm (5 percent to 10 percent). will decline by 20 percent and that road conditions will continue to deteriorate (figure E.7). Although it is 17. Regional allocation of resources. Current dis- important to use currently available resources more tribution of public resources may not be consistent wisely, road sector revenues must be increased and with underlying needs for road connectivity. Whereas diversified over the long run. allocation of maintenance resources is concentrated in Zambezia, Inhambane, Manica, and Sofala, it is esti- mated that Niassa has greater need, followed by Tete Figure E.7. Road Roughness Under Various Scenarios and Zambezia. 8 7 18. The way forward. To address these challenges, Average roughness (m/km) 6 the following actions are needed to restore and main- 5 tain road sector development (figure E.8). 4 • Strategic priorities must be established to avoid 3 unnecessary expenditures. Multisectoral strategic Optimal prioritization at the national level is needed based 2 Constraint 1 Constraint 2 on collaboration with other economic and social 1 Do minimum sectors (e.g., agriculture, health, education, private 0 0 5 10 sector development). The government has devel- Year oped a prioritization tool that should be improved and applied strategically to make road investments more selective because of the continued tight fiscal 1 00 /fair condition 80 11 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure E.8. Recommendations for Road Sector Development Figure E.9. Economic Flood Risk Expenditures • Strategic prioritization of investment • Efficiency in budget execution Administration • Efficiency in public procurement • Optimization of road expenditures Road asset management • Resilience in road infrastructure • Increase and diversification of revenues Fiscal sustainability • Mobilization of private resources space, growing uncertainty in the global economy, and unforeseen external shocks. • To use available resources wisely, the government must increase efficiency in budget execution. The budget execution rate is currently approximately 40 percent to 60 percent. A medium-term budget- ary framework, with more realistic project plan- ning, especially for donor-funded projects, must be developed. • A more solid, systematic project assessment and monitoring system must be built following an ongo- ing high-level macro-fiscal management improve- ment initiative, Gestao Economica para Desenolvi- mento Inclusivo. As the discovery of nonperforming external public loans in 2016 revealed, the govern- • Within the road-sector envelope, the government ment’s project assessment and governance systems should allocate more resources to road mainte- must be strengthened. nance—ideally, approximately 70 percent of total resources—while concentrating rehabilitation and • It may be possible to lower unit rates of road upgrading activities in top-priority primary roads. projects by strengthening market competition and encouraging local contractors to participate • Climate resilience should be built into road infra- more in public road procurement opportunities. It structure (figure E.9). Upfront investment could is important to achieve greater efficiency gains in result in higher net present value of road assets, public procurement by introducing new contractual even if average road conditions are still worse than methods, such as output- and performance-based under a no-climate-change scenario. road contracts and community-based maintenance contracts. It is expected that these mechanisms • Road-sector revenues must be increased, particu- will not only contain public procurement costs, but larly fuel levies, which have been fixed for a long also generate local jobs and facilitate local business time and must be kept adjusted, possibly in line with development. a regional norm (10 U.S. cents per liter) (figure E.10). 12 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure E.10. Fuel Levies in African Countries Figure E.11. Current Road Traffic 20 Fuel levy on petrol ($ Recommended = 17 cents 15 cents/liter) 10 Avg. = 8.9 cents 5 0 Be e Bu n in biq l i T a Cô ame n da K so nz i Gh ep . oz M e e ad Ch . M i bia Le i a an al d’l on S ie Eth ogo o, r ag ad Ga ea-B inea in u r Za di Co N b ia Na eny a C w a ia Af S en ire i a, u p Ta al aw ng i ge Gu G sca am a Th u an on mb i ssa ric eg R an Re rr a i op run Fa te r o vo R m m a na rk i M Bu M ral nt Ce 10000 levies should be dedicated more to road sector • Fuel development. Fuel levies are currently only partly RF resources ($/km) 8000 ring-fenced, undermining the principles of road 6000 funds, such as independence and sustainability. 4000 • Toll-based 2000 public-private partnership (PPP) oppor- tunities 0 may be limited, but other types of arrange- ments could be explored where traffic is expected S ie B al i a da M a so M i Co Gu n e na ia M i a, i a Ta Togo Bu Nam on o, a Rw scar Ga Ethi n di Ca Ken a Za ia me y a Cô S en u e Le n Gh ire ag e Ch . d’l al Bu ger Ni d w p i ng ine an rr a n i a ad Th rk i i b an mb op te eg mb Re al a o an vo o iq ru e F r a nz mb to grow. Mozambique’s readiness to implement a oz a road restoration window could be investigated. M Such a program would start with enhanced rev- enues from fuel taxes as a prerequisite. Then selected priority alignments would be restored using long-term performance-based road contracts, potentially leveraging private sector participation, Source: Based on ANE data. depending on traffic (figure E.11). Revenues from this charging system would be allocated to a road fund, which would fund availability and perfor- mance payments of road restoration contracts. Sur- plus revenues would cross-subsidize existing road programs, increasing revenues available to the road fund while gradually improving and maintain- ing road conditions. 13 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 1. Introduction 14 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 19. Until recently, Mozambique had achieved Figure 1. Contribution to Economic robust economic growth, mainly driven by the Growth of Various Sectors traditional extractive industries and the emerging 18 Service secgtor service sector. The country has economic potential 16 Non manufacturing industry Manufacturing in a wide range of sectors, such as agriculture, agro- 14 Agriculture Growth contribution (%) business, light manufacturing, and mining. During the 12 10 first half of the 2010s, the average annual growth rate 8 was approximately 7 percent, with a particularly strong 6 growth rate of more than 10 percent in nonmanu- 4 facturing industry and 9 percent in the service sector 2 (figure 1). Marked growth was also attributed to strong 0 1999 2000 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2001 performance in the mining sector. Coal production increased from approximately 30,000 tons in 2009 to 6 Source: WDI (2021) million tons in 2014, and its international market price doubled during the same period (INE 2015). In 2018, Mozambique also started producing liquefied natural the region, largely depending on land-intensive sub- gas along the northern coastline, with proven reserves sistence farming. More than 98 percent of farmed of an estimated 3.7 trillion m3, the second largest gas land is rainfed in Mozambique. With limited access reserves in Sub-Saharan Africa (after Nigeria). More to advanced inputs and technologies, agricultural than USD60 billion of foreign investment is expected in productivity is limited (Benfica et al. 2014; Benson, gas projects in the next decade (World Bank 2020a). Mogues, and Woldeyohannes 2014; Pauw et al. 2012). 20. Despite such strong resource-driven growth, 21. In addition, the recent debt crisis revealed the Mozambique has long been facing structural chal- country’s continuing challenge of governance, lenges to sustainable growth. The economy is highly especially in the public sector. In 2016, approxi- vulnerable to international commodity prices and mately USD2 billion (15 percent of GDP) of hidden economic conditions in trading partners. Coal prices nonperforming public debt was discovered, which halved from their peak in the early 2010s, and growth triggered a sharp currency depreciation from 50 to 80 of coal production reached a temporary plateau Mozambican meticals per U.S. dollar, increased the during the mid-2010s. Agricultural growth has also debt ratio from 50 percent to more than 100 percent been generally weak, fluctuating from 2 percent to 10 of GDP, and slowed the country’s economic growth percent depending on weather conditions. Agriculture substantially (e.g., IESE 2016; World Bank 2020a; World is an important economic sector, employing approx- Bank 2020b). In Mozambique, corruption and informal imately 80 percent of the workforce and generating sector practices remain among the most crucial busi- approximately 30 percent of gross domestic product ness constraints that firms identified (figure 2). (GDP). The country produces not only food crops for domestic consumption such as cassava, bananas, and maize, but also export crops such as tobacco, sugar, and cotton, totaling USD3 billion of agricultural com- modities,2 but productivity is among the lowest in 2 Based on FAOSTAT. 100% 80% 60%15 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 40% 20% 0% 2007 2018 2007 2018 2007 2018 2007 2018 2007 2018 Transport Electricity Telecom Customs & Labor trade regulations regulations No obstacle obstacle MinorConstraints Moderate obstacle Figure 2. Greatest Business Major obstacle Very sevrere obstacle According 23. Poverty has decreased substantially but to Firms, 2018 remains persistently high, particularly in rural Corruption areas, and regional inequality has widened. Access to finance Approximately 12.3 million people (48 percent) lived Practices of the informal sector Crime, thft and disorder below the poverty line in 2014/15—a marked drop Political instability Electricity from 58 percent in 2008/09, which was a marginal Tax rates Access to land reduction from the previous 6-year period (60 percent Customs and trade regulations Transportation in 2002/03) (World Bank 2020d), although inequal- Business licenses and permits Inadequately educated workforce ity has increased. Poverty is particularly prevalent in Labor regulations Tax administration the northern and central regions, where private sec- Courts tor investments have been modest. One reason for 0 5 10 15 Share of firms that identified as the biggest obstacle (%) the increasing inequality is the skewed distribution Source: Enterprise Surveys. of benefits from the emerging nonsubsistence sec- tors such as mining (Gradin and Tarp 2019). Although some resource-dependent growth, led by liquefied 22. Mozambique is also highly vulnerable to natu- natural gas projects, is expected in the short term, it ral disasters and other external shocks. Cyclones cannot support sustained, inclusive growth. Another Idai and Kenneth hit the country consecutively in 2019, may be misallocation of public infrastructure invest- causing devastating damage and adding challenges ment toward urban areas (World Bank 2019a). Many to the economy. According to a postdisaster needs rural farmers in remote areas have been left behind, assessment, the total cost of emergency and recon- for example in central and northern Mozambique.4 It struction projects was approximately USD1.5 billion is of particular importance to connect them to markets (10 percent of GDP) (IMF 2019). The current COVID-19 where formal businesses are concentrated because crisis is an additional challenge to Mozambique and job opportunities other than in agriculture are limited the global economy. More than 60,000 cases have in rural areas. been reported, and more than 700 people have died.3 Although a state of emergency was lifted to move to a 24. Inclusive growth with local business develop- state of public calamity in September 2020, uncertainty ment is necessary for sustainable growth. Because remains about the path of economic recovery from of poor transport connectivity and fragmentation the crisis. In Mozambique, the informal sector is thick, of domestic markets, Mozambique, with more than accounting for 90 percent of enterprises, but gener- 31 million population, has not taken full advantage ates only 30 percent of GDP (World Bank 2020a). There of the potential of its internal market. It has one of are many low-skilled, self-employed, or household the youngest populations in the world, with a rapidly enterprises in urban and rural areas. As of June 2020, increasing youth population. Approximately 45 per- 67 percent of the urban population was not working, cent of the population is younger than 15, and the approximately 120,000 jobs had been lost, and 63,000 total population is growing by 2.9 percent every year employment contracts had been suspended (World (World Bank 2021). Because the exhaustible, capi- Bank 2020c). Many self-employed individuals in the tal-intensive extractive industry and the unproductive informal sector are experiencing income loss and lack traditional sectors cannot generate sufficient jobs to access to social protections (World Bank forthcoming). absorb the rapidly growing labor force, informality 3 As of March 10, 2021, according to https://mz.usembassy.gov/ 4 There are different small-area estimates of poverty rates. See the covid-19-information/. following section for further details. 16 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING of the economy and regional inequality are rising. It economy. With increased emergency spending and has been a challenge to generate better-paying jobs slowed economic activities, an already difficult fiscal to employ young workers entering the labor force— position could become even tighter without proper approximately 500,000 per year. The unemploy- consolidation measures (IMF 2019a). ment rate for urban youth is 21 percent (World Bank forthcoming). 27. Given limited available resources and chron- ically constrained fiscal space, the efficiency and 25. Efficient, reliable transport connectivity is effectiveness of road spending must be increased among the most important conditions for sustain- by establishing a more sustainable mechanism able, inclusive growth. Greater transport connectivity to finance road investment and maintenance. can decrease transport costs and time, facilitating effi- The vast majority of Mozambique’s rural population cient mobility and creating more jobs and thereby alle- does not have good access to a reliable road network. viating poverty. In general, in the short term, greater Unmet financial need for universal access is estimated transport connectivity reduces transport costs and at approximately USD11 billion—nearly 80 percent of travel time (e.g., Danida 2010; Lokshin and Yemtsov Mozambique’s GDP. By African standards, the govern- 2005) and increases access to facilities such as schools ment of Mozambique has been relatively successful in and hospitals (e.g., Khandker et al. 2009; Mu and van mobilizing road user charges and has spent consider- de Walle 2011). In the longer term, agricultural pro- able resources on road sector development, although ductivity increases, firms become more profitable, there are many ways of increasing and diversifying and jobs are created (e.g., Bell and Van Dillen, 2012), road revenues and increasing efficiency in public all of which alleviate poverty (Dercon et al., 2008).5 spending (e.g., World Bank 2019b). In Mozambique, though, little evidence is available to support such linkages between transport connectivity 28. Against this background, the objectives of the cur- and economic outcomes. rent report are to update the status of the road net- work and recent developments in connectivity, exam- 26. In Mozambique, sustainable, resilient trans- ine some major socioeconomic benefits that have port connectivity is needed more than ever. Recent accrued from greater connectivity, review how the climate shocks that hit Mozambique highlight that government has been spending available resources on the country is highly vulnerable to extreme climate road sector development in recent years, and explore events. Cyclones Idai and Kenneth, which hit in March the potential to increase the efficiency and sustain- and April 2019, devastated the economy. Cyclone ability of road-sector spending. The report combines Idai brought strong winds and heavy rain, causing a variety of global and country data, including official river overflows, a storm surge, and floods over an statistics, micro surveys, and geographic open-source estimated 3,000 km2 of land and 715,378 hectares of data, to identify potential economic opportunities and cultivated land (UNDP 2019). Public infrastructure was bottlenecks across sectors. Box 1 outlines the limita- also severely damaged. The two cyclones damaged tions of the report. 3,600 km of roads, causing substantial disruptions in the transport system and therefore through the entire 5 See appendix A for a detailed literature review. 17 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Box 1. Limitations of this Report Important caveats should be considered when reading this report. First, data may not be up to date or may lack granularity. There is often a time lag for availability of transport data, such as road conditions and road user costs. A number of assumptions and approximations were made to fill the gap. The report relies on data that were available when the work was con- ducted (2010-18) and may not have completely captured the effects of more recent events such as cyclones Idai and Kenneth and COVID-19. Second, the objective of the report was to highlight a wide range of economic benefits from greater transport connectivity in Mozambique, but it covers only benefits of ground transport. For instance, transport externalities, such as air pollution and safety, are outside the scope of the report. There are a wide range of analyses and studies focused on individual issues in each sector, such as agricultural growth, job creation, and fiscal sustainability (e.g., World Bank 2020a; 2020b; forthcoming). Final policy decisions must take all this into account. Finally, related to the above, the report relies on micro analyses that take a partial equilibrium approach. The report provides a general sense of the direction and magnitude of the individ- ual effects in a less dynamic sense than general equilibrium models. Although the individual analyses are reliable in a statistical sense, the overall effect is uncertain when all of the effects are combined. 18 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 2.  Greater Market Accessibility but Stagnant Rural Access 19 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 3. Road Network and Population Distribution Existing Road Network 29. Roads are the most important mode of trans- port in Mozambique. The country has a road network of approximately 29,800 km of classified roads, of which approximately 8,300 km (27 percent) are paved (figure 3). The road network broadly covers the entire country and plays an important role in ensuring con- nectivity and mobility in the economy. Road transport accounts for half of freight traffic (figure 4) and 98 per- cent of passenger traffic in Mozambique (figure 5). 30. Despite its good geographic coverage, the road network is narrowly defined and should be Sea transport Air Pipeline 4.6% 0.1% expanded. Roads are classified as primary, secondary, 5.6% tertiary, and vicinal according to Decree 50/2000 of December 21, 2000. Primary roads are mostly paved CFM North and connect provincial capitals and main ports. Sec- 1.1% ondary roads connect primary roads and important CFM Road Central ports, border posts, and economic poles. Tertiary and 50.1% 28.9% vicinal roads connect district centers and villages. CFM South 9.7% Figure 4. Modal Split of Freight Transport Figure 5. Modal Split of Passenger Transport Sea transport Air Railway Sea transport Pipeline 4.6% 0.1% 0.5% 0.0% 5.6% Air CFM North 1.4% 1.1% CFM Road Central 50.1% 28.9% Road 98.1% CFM South 9.7% Source: INE (2015) Source: INE (2015) Railway Sea transport 0.5% 0.0% Air 1.4% 20 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 31. The density of the official road network (3.8 of GDP) on road and bridge construction, rehabili- km per 100 km2 of land) is among the lowest in tation, and maintenance under the Integrated Road the region and compares unfavorably with those of Sector Program from 2010 to 2018 (table 1, figure 7). neighboring countries, for example, 11 km in Kenya Many resources were used for road rehabilitation and and 6 km in Tanzania. Even in Zambia, where popula- upgrading projects, especially during the early 2010s. tion density is lower than in Mozambique, road density More than 40 percent of total spending was on reha- is approximately 5 km per 100 km2 (figure 6). Mozam- bilitating and upgrading national roads, although bique faces substantial challenges in developing road substantial amounts were also spent on mainte- connectivity and is likely to need to classify currently nance of roads and bridges (~30 percent). One major unclassified 16 feeder roads to expand the network. Classified road investment has been in a north–south corridor, Road density (km per 100 km2) 14 Total network National Highway N1 (figure 8), with other complemen- 12 32. 10 In the late 2000s to the early 2010s, the govern- tary rehabilitation and upgrading projects in surround- 8 ment6 of Mozambique made significant efforts to ing areas in Zambezia and Tete provinces. The adaptive invest in the road network, particularly the pri- 4 program loans that the World Bank and other donors 2 mary0 road network. As will be discussed below in have supported financed USD186 million from 2001 to more detail (Sections IV and V), the government spent 2007 and USD325 million from 2007 to 2017. Mozambique Africa average Ethiopia Kenya Tanzania Zambia Sub-Saharan an average of USD400 million per year (2.5 percent Figure 8. Major Road Investments, 2010-18 Figure 6. Road and Population Density in Selected Countries 16 14.9 Road density 109 120 Pop density Population density (per km2) Road density (km per 100 km2) 14 90 100 12 maintenance 11.1 Routine 10 in 2018 (km) 64 80 8 51 6.1 60 6 Existing roads 5.0 4.6 38 (km) 23 3.8 40 4 2Financial needs 20 0 ($ mil) 0 Africa average Mozambique Kenya Tanzania Zambia Ethiopia Sub-Saharan Rehab/ Reconst needs (km) 0% 20% 40% 60% 80% 100% Zambezia Nampula Niassa Tete Sources: ANE for Mozambique; Gwilliam (2011) for other countries; WDI (2021). Inhambane Cabo Delgado Gaza Manica Sofala Maputo Figure 7. Road Sector Spending 800 4.0 (US$ million) 700 3.5 Road sector spending (US$ % of GDP 600 3.0 Percent of GDP (%) 500 2.5 million) 400 2.0 300 1.5 200 1.0 100 0.5 0 0.0 2010 2011 2012 2013 2014 2015 2016 2017 2018 Sources: Source: ANE. Sources: Source: ANE. 21 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 1. Road Sector Spending under the Integrated Road Sector Program (Million Meticals) 2010 2011 2012 2013 2014 2015 2016 2017 2018 Administration 688 522 541 748 742 739 929 942 1,054 Capacity building 157 193 175 114 86 106 182 172 45 Maintenance of roads and bridges 4,072 2,714 3,179 3,658 3,804 4,881 3,685 3,824 7,735 Construction or rehabilitation of bridges 433 3,041 425 1,242 767 936 1,158 204 303 Rehabilitation or upgrade of regional roads 741 680 677 432 670 516 384 581 329 Rehabilitation or upgrade of national roads 4,027 5,227 6,235 9,100 13,604 5,045 2,377 3,901 5,213 Project preparation 0 0 26 5 12 10 6 88 948 Road safety 93 60 189 63 62 47 22 20 26 Other projects 0 0 0 0 1,377 970 243 316 116 Public-private partnerships 0 0 0 0 0 4,209 1,732 1,644 584 Total (MZN million) 10,211 12,438 11,446 15,363 21,125 17,459 10,718 11,692 16,353 (USD million) 301 428 403 510 674 437 170 184 271 % of gross domestic product 2.7 3.0 2.5 3.0 3.8 2.7 1.4 1.4 1.8 Source: Administracao Nacional de Estradas (ANE). 33. Primary paved roads were relatively well main- 34. Unpaved roads have long been neglected and tained before the cyclones hit in 2019, but second- are in poor condition. There are approximately ary and tertiary roads were and remain in poor con- 21,600 km of unpaved roads, of which approximately dition. Because of intensive government investment 44 percent were in poor condition in 2017, up from 42 in upgrading projects, the total length of paved roads percent in 2010. These poorly maintained rural roads increased by nearly 30 percent, from approximately are highly vulnerable to climate events, such as heavy 6,100 km in 2010 to 8,100 km in 2017 (figure 9). Paved rainfall. Thus, one clear challenge is that, given limited roads are kept in relatively good condition. According available resources in the road sector, how can the to georeferenced road data from Administracao Nacio- country keep the primary paved network well main- nal de Estradas (ANE), in 2010, about half of paved tained and improve the unpaved rural road network? roads were in good condition, and the other half were in fair condition. By 2017, about 91 percent of paved roads were in good or fair condition, and the rest were in poor condition, indicating that some maintenance backlogs may have accumulated during the late 2010s (figure 10). The two cyclones destroyed approximately 2,000 km of roads, 142 culverts, and 29 bridges and damaged 35 drifts, leaving 6,142 km of roads (~20 per- cent of the total road network) impassable. 12000 11833 22 10000 8000 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 6000 4676 5375 4000 4000 2000 1287 863 542 57 0 Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved Primary Secondary Tertiary Vicinal Figure 9. Road Length (km) According to Condition, 2010 and 2017 Good Fair Poor (2010) (2017) 12000 11833 12000 11483 10000 10000 8000 8000 6200 4676 5375 5661 6000 6000 4000 3440 4000 4000 1287 863 1346 949 2000 542 57 2000 532 182 0 100%0 0 0 7 17 80% 37 39 Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved 46 60% 68 40% 20% 51 50 47 Primary Secondary Tertiary Vicinal Primary Secondary Tertiary 16Vicinal 1 0% 7 9 7 Good Fair Poor Good Fair Poor Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved Source: ANE 12000 11483 10000 Primary Secondary Tertiary Vicinal 8000 6200 6000 10. Share of Road According to Condition, 2010 and 2017 Figure 5661 Good Fair Poor 4000 3440 1346 (2010) 949 (2017) 2000 532 182 100%0 0 7 100% 0 17 6 3 26 80% 37 39 80% 40 39 Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved 46 59 54 60% 68 60% 72 40% 40% 20% 51 50 47 20% 52 49 Primary Secondary Tertiary 37 7 9 7 16Vicinal 1 0% 7 7 5 0% 0 1 Good Fair Poor Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved Primary Secondary Tertiary Vicinal Primary Secondary Tertiary Vicinal Good Fair Poor Good Fair Poor Source: ANE 100% 6 3 26 80% 40 39 59 54 60% 72 40% 35. 20% 52 is a marked There 49 difference in road condi- south. There is similar regional inequality in second- 37 0% between 0 tion provinces, particularly 7 7between 5 the 1 ary and tertiary roads. Climate events intensifies this north and the south. Although the primary network regional heterogeneity. Damage from the cyclones Paved Paved Paved Paved Unpaved Unpaved Unpaved Unpaved is generally relatively well maintained, the same is not was concentrated in the northern provinces, such as true in the provinces, such northern Secondary Primary as Cabo Delgado, Tertiary Vicinal Sofala, Tete, and Zambezia, where road conditions are Nampula, Niassa, and Zambezia. Good Fair Forty-two Poor percent of worse. This aggravates recently widening inequalities primary roads are in poor condition in Niassa, 21 per- in the country and requires a more strategic realloca- cent in Zambezia, and 19 percent in Cabo Delgado (fig- tion of public resources to those provinces for more ure 11). The primary network carries the vast majority inclusive, sustainable growth. of the nation’s traffic and connects the north and the Share of roads (%) Shaer of roads (%) 23 Ca Ca bo bo 100 100 10 20 30 40 50 60 70 80 90 10 20 30 40 50 60 70 80 90 0 0 De De l ga l ga Source: ANE do do 10 46 26 19 Ga Ga In za In za 0 5 72 12 ha ha m m ba ba ne 0 ne 73 13 29 M M an an ica 0 Good ica 89 Good 16 24 M M ap ap ut ut 0 2 95 11 Fair Na o Fair Na o m m (Tertiary) (Primary) pu pu la la 8 32 11 38 Poor Ni Poor Ni as as sa 6 42 sa 13 51 So So fa fa la 0 la 34 3 38 Te Te 0 23 te Za te 16 54 Za m m be be z zia 54 21 ia 0 49 Figure 11. Road Condition According to Province and Type of Road Share of roads (%) Ca bo 100 10 20 30 40 50 60 70 80 90 0 1000000 1200000 1400000 1600000 200000 400000 600000 800000 0 Ma De put l ga Ma Nam to oa do 47 17 pulla Bei r Chim a Ga o In za 19 20 Que Nacaio ha li ma l a Mo ne m c ub ba ne 42 35 T ta Gue e M L ich rue an inga 1997 Pem ica Good 31 12 X a ba M M i -Xai ap Anga xi xe ut oc 0 0 Fair MoCuamhe 2007 Na o nte ba pue m Inha Do nd z pu (Secondary) la 0 69 mba o Chib ne Poor Ma uto Ni n as sa Mo Chóhi ca 0 52 çam kwè biqu 2017 (est.) Nam e So Al to Mon e ti fa la 9 16 Mo Mo apol cím Ma r l ocu bo a ro m e Te da eu 27 29 MoPraia Za te Vi la ati ze n m Maculo be zia 0 Go n ni ca 46 do l a Chiu Nam re TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING INVESTING IN MOZAMBIQUE’S CONNECTIVITY: Mi la i alo nge 24 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Market Accessibility 36. There is marked variation in domestic market proximity to consumption areas. By contrast, some accessibility in Mozambique. In general, improved inland areas in Manica, Niassa, and Tete provinces, roads increase accessibility to markets because bet- where transport costs exceed USD30 per ton, are dis- ter road surfaces are more reliable and allow people advantaged in terms of domestic market accessibility. to move faster at lower cost, saving on transportation cost and time. The expected effect of road investment 37. The global market is also far from some of the is often not linear but varies according to location. inland areas. Ports can be considered as a proxy for Using road condition data and road user charge esti- global and regional markets. There are three major mates, the network was analyzed to estimate transport ports in Mozambique: Maputo, Beira, and Nacala. costs. Large cities are important markets, for exam- Along the major corridors to these ports, global mar- ple, for agricultural produce, such as maize, cassava, ket access is relatively good, but Niassa, Tete, and Gaza and other food crops. The estimated cost of bringing provinces are disconnected from the ports, and hence 1 ton of goods to a city with more than 50,000 inhab- global markets, despite recent major road improve- itants varies substantially according to region (figure ments (figure 13). Additional efforts are needed to con- 12). Peri-urban areas have the natural advantage of nect these areas more closely to the global market. Figure 12. Transport Costs to a Large City (USD/Ton) Figure 13. Transport Costs to a Major Port (USD/Ton) Source: World Bank estimate. Source: World Bank estimate. 25 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 14. Market Access Index, 2010 and 2017 (2010) (2017) Source: World Bank estimate. Note: 2017 data were normalized based on maximum value of the 2010 market access index. 38. Domestic market accessibility is important for 39. From 2010 to 2017, intensive investments in pri- mitigation of regional inequalities and promotion mary roads such as National Highway N1 substan- of more inclusive growth. In general, the market tially increased domestic market accessibility (fig- accessibility increases with the size of purchasing ure 14). Accelerating urbanization is another contrib- power or market capacity and decreases with the utor to this improvement: By definition, MAI increases degree of impediment between origin and destina- with share of urban population, holding everything tion. The Market Access Index (MAI) is often defined else constant. From 2007 to 2017, large Mozambican as the sum of purchasing power or market capacity, cities became even larger (figure 15), with the total y, inversely weighted by the degree of impediment, d, population of the 10 largest cities more than doubling between two locations, i and k:6 from approximately 900,000 to 1.9 million.7 Altogether, market accessibility increased substantially for many Mozambicans who live near large cities and along main corridors, although some areas, such as Niassa, Tete, and Gaza, still lack market accessibility. 6 This is based on a conventional gravity framework. The literature also uses negative exponential weights. See for instance Elbadawi, Mengistae, and Zeufack (2006) and Lall and Mengistae (2005). 7 Maputo, Matola, Nampula, Beira, Chimoio, Nacala, Quelimane, Mocuba, Tete, Gurue. 40 30 Share 20 47 42 26 10 19 31 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: 27 TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 0 9 0 0 0 0 zia Na o za do Za te sa ica la la ne ut fa pu Ga Te as ga be ba an ap So Ni m l m m M De M ha bo Good Fair Poor In Ca Figure 15. Population According to City, 1997, 2007, 2017 1600000 1997 2007 2017 (est.) 1400000 1200000 1000000 800000 600000 400000 200000 0 Anga xi xe li ma l a e Xai - ba e a Al to on e ti l bo a ro m e ni ca Vi la ati ze nge Mo ne Ch n e ei ra MoPraia T ta nga MoCuamhe L ich rue z do l a Ma ibuto Mo Chóhi ca Nam re M Xai pulla da eu cím Ma rol ocuo çam kwè Mi la i alo mba o nte ba to o Que acaio Maculo pue c ub biqu Inha Do nd put M p o Gue Chiu Pem oc a Nam Chim i B Ma n n Go n Nam Ma N M Mo Source: Based on 1997, 2007 and 2017 (preliminary) census data from the Instituto Nacional de Estatistica. 40. The north–south corridor (N1) is increasing to inclusive growth, benefiting more Mozambican domestic market accessibility, connecting the people. Measured by the number of people who could northern and southern parts of the country. Differ- benefit from improved market accessibility (with a ent corridors contribute to different types of accessibil- threshold of 2 percentage points), the full north-south ity. To compare their effects, a simple simulation was corridor development could 4 times more inclusive conducted using the same geographic data as above. than the three east-west corridor developments. While Four national highways were considered: a north– 56 percent of the total population could benefit from south corridor (N1) and the Maputo (N4), Beira (N6), the north-south, the east-west corridors could affect and Nacala (N12 and N13) corridors.8 Because these only 14 percent mainly in the northern provinces. major corridors were already in relatively good condi- tion except for some sections of N13, it was assumed 42. Investment needs and expected benefits are that the road conditions would be improved in a high different in the different corridors. Whereas N1 range of roughness.9 The results are in sharp contrast is longer than 2,400 km, the east–west corridors are between the north-south corridor and the east-west much shorter (N4, 90 km; N6, 280 km; N12, 105 km; corridors (figure 16). Although potential improvement N13, 640 km). In addition, the role of regional cor- of N1 could increase market accessibility throughout ridors in connecting to neighboring countries was the country, the expected effect of improving the three not taken into account. According to this simulation, east–west corridors is less and more specific to certain further improvement along the Nacala corridor could areas, such as Niassa and Nampula, where road con- help open up Niassa, but in the context of the national ditions are still poor. The lengths of the corridors (and market, the north–south corridor would have greater thus the sizes of necessary investments) are different. effect because Greater Maputo, including Maputo and Matola, is the primary city in Mozambique. Thus, for 41. It is difficult to compare expected benefits from domestic market integration and local business devel- the two different scenarios, however, it is clear that the opment, it is essential to improve and maintain the north-south corridor development is more conducive north–south corridor. 8 For simplicity, the analysis ignores a section where the Nacala corridor overlaps with N1. 9 It was assumed that the international roughness index would be improved from 2017 conditions to 1.5, which means that a road is in very good condition. 27 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 16. Estimated Changes in Market Access Index (%) North–South Corridor East–West Corridors Source: World Bank estimate. Rural Accessibility 43. Rural connectivity has long been a challenge in (World Bank 2016). The RAI is calculated according to Mozambique. An estimated 16 million people did not the proportion of people with access to an all-season have access to the good road network in 2017, which road within an approximate walking distance of 2 km results in a rural access index (RAI) of 19.3 percent. The or a walking time of 25 minutes. index is one of the most important global indicators in the transport sector and is used for Sustainable Devel- 44. Rural accessibility has deteriorated in Mozam- opment Goal Target 9.1: to “develop quality, reliable, bique in recent years. The RAI was initially estimated sustainable and resilient infrastructure…to support at 27 percent in 2006 and declined to 20.4 percent in economic development and human well-being, with 2010, although there is a methodological difference a focus on affordable and equitable access for all” between the two figures (box 2).10 10 See a more detailed discussion in World Bank (2016). 28 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Box 2. Measuring Rural Accessibility Using New Technology The Rural Access Index (RAI) is one of the most important global indicators for use in assessing people’s connectivity in the transport sector. Roberts et al. (2006) developed the original RAI, which used household surveys, with some modeling when data were not available. It is defined as the share of people who have access to an all-season road within an approximate 2-km walking distance. Although how to measure it is not entirely clear, the original RAI estimates that, globally, 1 billion people (~31 percent of the rural population) do not have access to a road network. In Africa, it is estimated that 450 million people (more than 70 percent of the rural population) are unconnected. This compares unfavorably with many developing countries. For instance, the RAI is estimated to be 90 percent in East Asia and 58 percent in South Asia. Despite its important contribution to transport operations, the original RAI has several lim- itations. First, it is difficult to update data in a timely manner because household surveys are costly and are not available in all countries. Second, because it uses household surveys, geo- graphic representation is not perfect. The resultant index is only available at the national level. Finally, related to the above, it has little operational relevance. The estimated RAI cannot be used to infer accessibility at a given location. In addition, the original RAI may not reflect actual efforts to construct or rehabilitate roads because of its incomplete geographic representation. If the sample of a household survey does not cover a project area, the index does not change even though people’s accessibility surely increases. To overcome these disadvantages, a new RAI is being developed using new technologies, such as geo-referenced road network data and highly disaggregated population distribution data (e.g., WorldPop). With the quality of roads taken into account, the number of people living a within 2 km of the nearest road in good condition is estimated. This new RAI is expected to contribute to the ongoing Sustainable Development Goal discussion. For Mozambique, the original RAI was estimated to be 27 percent in 2006, which was inferred using a modeling approach, with no household survey used. The RAI was estimated to be 20.4 percent in 2010 and 19.3 percent in 2017—lower than the previous estimate because the new RAI takes the quality of roads into account more explicitly. An advantage of the new RAI is that it can measure accessibility at the highly disaggregated level, such as the district. It is useful for investment decisions. a A road in good condition is tentatively defined as a paved road in good or fair condition or with an international roughness index less than 5 or an unpaved road in good condition or with an international roughness index less than 10. 29 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 45. The current methodology explicitly accounts for Manhica, Bilene, and Maputo (45 percent). In most changes in road quality at the individual road level. rural areas, the RAI is estimated to be less than 10 With newer data for 2017, accessibility is estimated at percent (figure 18). Accessibility is particularly limited 19.3 percent, a reduction from 2010 (figure 17). The in the northern and inland provinces. In addition, the number of rural people who do not have access to the regional gap was widened. From 2010 to 2017, rural reliable road network increased from 14.9 million in accessibility improved in Maputo Province (including 2010 to 16.1 million in 2017. Maputo City), where the index is highest (44 percent) (figure 19), whereas it worsened in Nampula, Niassa, 46. Rural accessibility varies significantly according and Tete where accessibility was already low. This is to location, seemingly widening regional inequal- consistent with the regional inequality (e.g., World ity. Road accessibility around major urban areas is Bank 2019). To reduce the gap, further efforts are generally higher than in rural areas. At the district called for than before. level, the RAI is high in Chokwe (60.8 percent) and Figure 17. Rural Access Index, 2010 and 2017 (2010) (2017) Source: World Bank estimate. 20 16.1 13.414.5 15 8.3 Rural Acc 8.2 10 30 5 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 0 sa z ia la la ne Na to a do Za ete l M a ta ic In Gaz fa pu as u ba To ga be an T ap So Ni m l m m M De ha bo Ca Figure 18. Rural Access Index According to District, 2017 70 Rural Access Index, 2017 (%) 60 50 40 30 20 10 0 Funhalouro Muidumbe Mogovolas Cheringoma Mopeia Namacurra Sussundenga Quissanga Namapa Maravia Morrumbala Namuno Chigubo Mavago Zumbu Mongincual Mossuril Ile Chokwe Gile Milange Barue Matutuine Gurue Marracuene Moamba Changara Chibabava Macossa Chicualacuala Panda Nacala Velha Source: World Bank estimate. Figure 19. Rural Access Index According to Province, 2010 and 2017 45 43.6 44.1 36.9 RAI, 2010 40 Rural Access Index (%) 35.4 35 25.4 RAI, 2017 30 20.4 24.5 22.2 25 19.3 20.0 20.3 17.5 17.2 18.4 16.9 14.6 14.6 20 16.1 8.3 13.414.5 15 8.2 10 5 0 sa z ia la la ne a Na to do Za ete l M a ta ic In Gaz fa pu as u ba To ga be an T ap So Ni m l m m M De ha bo Ca Source: World Bank estimate. 70 Rural Access Index, 2017 (%) 47. The poor tend to 60 suffer more from limited road reduction stagnated, declining to 52 percent by 2007, 50 reduced poverty in the late accessibility. Mozambique although per capita GDP grew to USD338, indicating 1990s and early 2000s, 40with the percentage of people a possible widening of income inequality in the econ- 30 line declining from 69 per- below the national poverty omy. According to the latest poverty assessment, the 20 in 2002 and GDP per cap- cent in 1996 to 54 percent poverty rate dropped substantially, to 48 percent, in 10 increasing from USD196 ita (in 2005 constant terms) 2014/15. to USD272, a 38 percent 0 increase. After that, poverty Funhalouro Muidumbe Mogovolas Namacurra Cheringoma Mopeia Sussundenga Namapa Quissanga Maravia Morrumbala Namuno Chigubo Mavago Zumbu Mongincual Mossuril Chokwe Ile Milange Barue Matutuine Gile Gurue Marracuene Moamba Changara Chibabava Panda Macossa Chicualacuala Nacala Velha 31 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 48. Poverty is particularly persistent in rural areas and 30 percent in Gaza Province. Poverty is much and is concentrated in the central and northern parts higher in the northern and inland provinces: 67 per- of the country. To address poverty at the national cent in Tete, 66 percent in Zambezia, and 63 percent level, it may be of particular importance to support in Niassa. Although causality is complex, rural acces- those lagging areas. There are several different small- sibility is highly correlated with poverty (figure 21). area estimates of poverty, depending on the empirical Poverty rates tends to be high (more than 60 percent) model of household consumption used (Hussein and in the districts where rural access is less than 20 per- Castigo 2012; Simler and Nhate 2005). According to cent. This merely shows correlation, not causality, but World Bank estimates, the poverty rate varies from this type of correlation is commonly observed in other region to region (figure 20) and in general is lower in developing countries. It is likely that high poverty coin- the southern provinces. It is estimated to be 2 percent cides with poor road accessibility for good socioeco- in the city of Maputo, 8 percent in Maputo Province, nomic reason. Figure 20. Poverty Rate Estimates, 2007 Figure 21. Rural Access Index and Poverty Rate 100 Rural Access Index (%), 2010 80 60 40 20 0 0 .2 .4 .6 .8 Poverty Rate, 2007 Source: World Bank estimate. 32 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Section Summary 49. Key takeaways from this section are as follows: • Because of the government’s intensive investments in the last decade, the primary paved road network has improved substantially over the last 10 years, increasing people’s market accessibility in the coun- try, particularly along the main corridors. • The quality of the paved network is generally good, although it has deteriorated slightly in recent years, calling for more timely maintenance with finan- cial sustainability ensured. There is also significant regional inequality. The primary road network in the northern provinces, such as Cabo Delgado, Niassa, Nampula, and Zambezia, is in poor con- dition, which adds to the climate vulnerability of these regions. • The north–south corridor (N1) has a critical role to play in integrating domestic markets and pro- moting more inclusive growth. Different corridors provide different types of connectivity. Whereas the east–west corridors, such as Maputo, Beira, and Nacala, are important as regional corridors, the north–south corridor provides domestic market accessibility. • Rural accessibility is a more general challenge. Approximately 44 percent of unpaved roads are in poor condition, leaving many rural people discon- nected, particularly in Nampula, Niassa, and Tete. The RAI was almost constant from 2010 to 2017. Approximately 16 million people do not have access to the road network in rural Mozambique. • The poor suffer from lack of rural connectivity. Particular attention should be paid to promoting inclusive, sustainable growth by developing last- mile connectivity in rural areas, possibly including currently unclassified feeder roads. 33 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 3.  Wider Socioeconomic Benefits from Transport Connectivity 34 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 50. Although there is strong correlation between trans- high. As a result, greater transport connectivity and port connectivity and poverty, how the economy can better economic outcomes tend to coexist. benefit from better transport connectivity is open for discussion. The literature suggests that greater trans- 51. To examine and quantify the effects of road con- port access can bring a wide range of economic and nectivity on the economy, the following five factors social benefits in the short and long term (figure 22). A are assessed: agricultural production, firm production, more detailed discussion is provided in appendix A. An agglomeration economies, access to health facilities, important empirical concern is endogeneity. Although and resilience and criticality of the road network. The road investment can reduce transport costs and there- depth of analyses varies depending on the availability fore increase economic productivity, road agencies also of data. often invest in areas where productivity is inherently Figure 22. Results Chain Access to Access to Income and Physical Access to Local market education and input/ output welfare connectivity labor market development health markets increase 1-2 years 3-5 years 5-10 years Greater Access to Domestic Market Increases Agriculture Production 52. Despite rapid development in nonsubsis- cotton exports are USD200 million and USD40 million tence sectors such as mining, agriculture remains per year, respectively.11 Productivity is low, largely an important economic sector in Mozambique, because production is based on subsistence farming, employing approximately 80 percent of the coun- with few advanced inputs such as fertilizer, irrigation, try’s workforce and generating 30 percent of GDP. and improved seeds used. Cereal yield in Mozambique Mozambique produces approximately USD3 billion of averaged approximately 0.8 tons per hectare from agricultural crops. Domestic food crops such as cas- 1996 to 2016, which is approximately 10 percent to sava (USD1 billion), bananas (USD132 million), and 20 percent of the world average yield and among the maize (USD125 million) are predominant. In addition, lowest even in Africa (Zavale et al. 2020). Most poor Mozambique is a traditional exporter of cash crops people in rural areas depend on agriculture, with few such as tobacco, sugar, and cotton. Tobacco and other job options. 11 Based on FAOSTAT. 35 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 23. Agricultural Production Value (USD million) Figure 24. Agricultural Potential (USD million) Source: IFPRI SPAM Update 2010. Source: IFPRI SPAM Update 2010. 53. From the agro-ecological point of view, Mozam- or institutional constraints, such as protected land, bique has significant untapped agricultural poten- but the estimate indicates that current production is tial. Geographically, production areas are concen- far below potential. Greater agricultural productivity trated in Nampula and Zambezia provinces and coastal will have important economic implications for inclu- areas of the southern provinces (figure 23). According sive growth and poverty reduction. It is also expected to the Spatial Production Allocation Model (SPAM) to increase self-sufficiency in food and improve the developed by the International Food Policy Research balance of payments. Despite its domestic potential, Institute,12 the country has agricultural potential of Mozambique imports approximately USD1 billion to USD567 billion under conditions of high commodity USD1.5 billion of food and agricultural products annu- prices (figure 24). SPAM does not assume any physical ally (figure 25). 12 SPAM is a spatial model to allocate crop production derived from large statistics reporting units such as province and district to a raster grid at a spatial res- olution of approximately 10 km x10 km. Using a cross-entropy method (Shannon 1948), the model infers likely production locations of 42 crops under four production systems in 2010 (see appendix B for more details). 36 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 54. Summary statistics reveal that agricultural Figure 25. Mozambique: Agricultural Exports and Imports productivity decreases with market accessibility 1800 Agricultural exports 1600 measured according to transport costs to the near- Agricultural imports Agricultural exports/imports 1400 est city with a population of 50,000 or more (figure 1200 (US$ million) 26). For maize, for instance, average yield is estimated 1000 at 1.2 tons per hectare in districts with good access 800 600 to markets (transport cost of less than USD2), but 400 where transport cost exceeds USD20, maize produc- 200 tivity is nearly 20 percent lower. Similar results can be 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 obtained for cassava. Limited transport accessibility is a crucial constraint on agricultural production in rural Source: FAOSTAT. 10 Mozambique. 9 Average yield (MT/ha) 8 26. Average Crop Yields and Market Access Figure 55. Transport infrastructure is not the only con- 7 (Maize) straint that farmers face. Only 10 percent of arable 6 land is in use, of which 90 percent is used for small- 5 scale subsistence farming. Low fertilizer adoption 4 is partly attributed to high transportation costs for <2 2~5 5~10 > 10 Transport cost to a city with 50,000 population (US$/ton) farmers to access input and output markets (Zavale et al. 2020). The government development strategy is comprehensive, aimed at improving rural infrastruc- ture, including road networks and storage facilities. It envisages doubling productivity in the sector. The expansion of year-round road access to areas with high productive potential is identified as a key strategy to achieve a annual agricultural growth of 7 percent Source: World Bank calculation based on Spatial Production (Ministry of Agriculture 2010). Allocation Model data. (Cassava) 56. To investigate the effects of transport connectivity in more detail, a translog agriculture production func- tion is estimated using agricultural sample surveys from 2012 and 2015, each of which covered approxi- mately 6,200 households in approximately 350 admin- istrative posts nationwide:13 ' ' ln #!"# = %$ + %%& ln '("# + )!"# %( + *!"# %) + +" + +# + ,!"# where vij is the total value of crops that household i in administrative post j produced at time t. All crops are Source: World Bank calculation based on Ministry of Agriculture 2012 aggregated using local or FAOSTAT crop prices. A set agriculture survey. 13 See a full discussion in Iimi 2021a. 37 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING of input variables (X) is considered: labor (L), land area Figure 27. Mozambique: Agricultural Exports and Imports (H), fertilizer use (F), pesticide use (P), and outside labor employed (N). Transport connectivity (TR) is measured 0.6 Marginal effect 0.5 +1.64*S.E. according to three variables: transport costs to the Marginal effect 0.4 -1.64*S.E. 0.3 nearest city, RAI, and transport costs to the nearest 0.2 0.1 port. Z is a set of household characteristics. The speci- 0.0 -0.1 fication generally follows the literature (e.g., Bravo- -0.2 Ortega and Lederman 2004), but lack of availability of e e nd ing I or r t t er RA bo or ke ci d vic li z l ab La ai n op ar La er sti rti data constrains it. Although some important produc- om de ns Tr t Pe Fe ts tsi si o t s co ts Ou tion factors, such as access to extension services, are en s TR co t Ex TR included in Z, other factors are not available, such as Source: World Bank estimate. access to storage facilities. Note: S.E., standard error. 60000 Number of firms registered 50000 57. Two approaches are used to address the endoge- 40000 neity associated with infrastructure placement. First, to agricultural production.15 Apart from transport 30000 the highly disaggregated location-specific fixed effects connectivity, the elasticity of land input is found to be 20000 (cj) are included. The dataset used is not a panel but significantly high, reflecting that traditional agricultural is composed of two rounds of cross-sectional data at production 10000 in Mozambique is focused on extensifica- two points in time, although the administrative posts tion rather 0 than intensification. The elasticity of labor 2010 2011 2012 2013 2014 2015 2016 are considered to be small enough to control for input is low, indicating that labor productivity is low location-specific unobservables such as agro-climatic and that labor is too abundant in rural areas. Not only potential. Second, the instrumental variable technique road infrastructure, but also other complementary is used with two instruments: straight line distance policies, such as extension services and distribution of from the nearest historic port developed in the 1860s advanced inputs, are important to stimulate agricul- and distance to rail lines developed in the colonial tural growth. era. These are considered to be exogenous to a cer- tain extent, because these historical events happened 59. To stimulate agricultural growth, further based mostly on political motivation.14 investments may be needed, for instance, in Inhambane, Niassa, Tete, and Manica, where there 58. A 10 percent reduction in transport costs can is untapped agricultural potential, but transport increase agricultural production value by 0.7 per- costs to markets are high. The result is support- cent to 1.1 percent, depending on specification. ive of the view that road investments could help link The main results are summarized in figure 27. Among potential agricultural areas, as envisaged in the gov- the transport connectivity variables, accessibility to ernment’s Integrated Road Sector Program. The result domestic markets was found to be most important. indicates that greater market accessibility would help Rural accessibility may also be important. When the increase agricultural production. instrumental variable regression is applied, the elastic- ity associated with RAI is found to be significantly pos- itive, although port access seems to be less relevant 14 According to conventional statistical tests, such as the Hausman exoge- 15 This may be because export crops still account for a small portion of total neity test and the over-identifying restriction test, the proposed instru- agricultural production in Mozambique. In particular, the agricultural ments were found to be valid. sample surveys used focus on small farmers. 0.6 Marginal effect 0.5 +1.64*S.E. Marginal effect 0.4 -1.64*S.E. 38 0.3 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 0.2 0.1 0.0 -0.1 Better Access to Ports Augments Firm -0.2 e e nd ing I or r t t er RA bo or ke ci d vic li z Productivity and Agglomeration Economies l ab La ai n op ar La er sti rti om de ns Tr t Pe Fe ts tsi si o t s co ts Ou en s TR co t Ex TR 60. Better transport infrastructure and thus low Figure 28. Number of Firms Registered transport costs can stimulate manufacturing 60000 production. Good-quality transport infrastructure Number of firms registered 50000 is essential to increase manufacturing productivity, 40000 potentially generating more jobs (e.g., Holl 2004; Limão and Venables 2001). Particularly with good 30000 transport connectivity, firms can minimize their inven- 20000 tory costs (Shirley and Winston 2004). 10000 61. In Mozambique, the formal manufacturing 0 2010 2011 2012 2013 2014 2015 2016 and service sectors are growing but remain gen- erally thin. From 2014 to 2016, the formal business Source: INE Indicadores económico-financeiros das. sector grew rapidly, and before the crisis in 2016, approximately 58,000 firms were registered (figure Figure 29. Share of Registered Firms According to Province 28). Firm distribution has historically been concen- Zambézia Tete Cabo trated in major urban areas. The City of Maputo and 4.1% Inhambane 3.1% delgado Maputo Province contain approximately half of total 4.9% 3.0% enterprises, followed by Gaza, Manica, and Niassa (figure 29). The northern provinces generally have Sofala fewer enterprises, meaning that few job opportunities 5.3% Maputo Cidade Agric are available to people living in those areas, except in Nampula 40.1% 3 agriculture. The service sector dominates formal busi- 5.8% nesses, accounting for 83 percent (figure 30). Manufac- Niassa turing is thin, and there are some agribusinesses. 7.5% Gaza 62. The manufacturing sector has been experienc- Manica 8.4% Maputo ing a weakening trend of technical capacity and 7.7% 10.1% skills base (Warren-Rodrigues 2010). Several techno- Source: INE Indicadores económico-financeiros das. logically capable firms established during the colonial period have closed since independence. Despite past 100% and resurging foreign direct investment in the country, 90% accumulation of skills and technical capability remains 80% weak, because large-scale foreign investments are 70% 60% often made by foreign companies, particularly in the 50% extractive industries, which prevents a trickle-down 40% 30% effect on the domestic economy. In addition, most 20% manufacturing companies are small, with fewer than 10% 10 employees. Thus, firm productivity has not improved 0% Transport Electricity Telecom Custom much in recent years (Ministry of Planning and Devel- trade opment 2013). Informality of the Mozambican economy regulat is persistently high (World Bank 2020a; forthcoming). No Obstacle Minor Obstacle Mode Major Obstacle Very Sev ere Obstacle 39 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 63. Many firms identified transport as a major or to the perceptions of business managers. Accessi- severe obstacle. Together with electricity, transport bility to markets and ports is poor in regions other has been one of the most significant business con- than Greater Maputo. Excluding Maputo and Matola, straints in the country (figure 31). It appears that the although approximately 80 percent of registered firms transportation constraint was resolved to a certain are located within 2 hours of a major domestic mar- extent between 2007 and 2018, whereas the energy ket (figure 32), only 30 percent are within 2 hours of a constraint appears to have become worse, according regional port (figure 33). Figure 30. Share of Registered Firms According to Sector Figure 31. Business Constraints bo Agro- Other Hotel 100% ado business manufactu 3% 0% 5% ring 80% Textile 2% 4% 60% Mining 100% 0% 40% Agriculture 80% 20% 3% 60% 0% 2007 2018 2007 2018 2007 2018 2007 2018 2007 2018 40% Transport Electricity Telecom Customs & Labor 20% Other trade regulations 0% services regulations No obstacle Minor obstacle Moderate obstacle Maputo 83% 2007 2018 2007 2018 2007 2018 2007 2018 2007 2018 Major obstacle Very sevrere obstacle 10.1% Transport Electricity Telecom Customs & Labor Source: INE Estatisticas Territoriais 2014. trade regulations Source: Mozambique BEEP 2007, 2018. regulations No obstacle Minor obstacle Moderate obstacle 100% Major obstacle Very sevrere obstacle 7.8 13.0 2.9 registered firms 80% 10.2 4.9 Figure 32. Share of Firms with Market Access Figure 33. Share of Firms 7.0 with Port Access 17.0 100% 60% 100% 11.7 7.8 13.0 2.9 19.5 Share of registered firms Share offirms 80% 10.2 4.9 40% 80% 32.4 7.0 17.0 72.0 10.0 53.4 Share of registered 60% 20% 60% 10.4 11.7 7.9 16.7 40% 0% 40% 17.3 ity Telecom Customs & 72.0 Labor trade regulations 53.4 All districts 52.2 Maputo/Matola Exclude13.2 20% regulations 20% < 30 min 30 min - 1 hour 1-2 2 - 3 20.4> 3 hours Minor Obstacle 0% Moderate Obstacle 0% Very Sev ere Obstacle All districts Exclude Maputo/Matola All districts Exclude Maputo/Matola < 30 min 30 min - 1 hour 1-2 2-3 > 3 hours < 1 hour 1-2 2-4 4-6 > 6 hours Source: World Bank estimates. Source: World Bank estimates. 0.4 Marginal effect 0.2 +1.64*S.E. Output elasticity -1.64*S.E. 0.0 -0.2 -0.4 -0.6 -0.8 Local market Inventory Agglomeration interruption MAI Wage Power outage Input goods Transport costs Electricity economies price prices Water access to port 40 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 100% 19.5 Share of registered firms 80% 32.4 10.0 60% 10.4 7.9 16.7 40% 17.3 64. To examine the possible effect of transport the elasticity of port 52.2 13.2 meaning accessibility was –0.079, 20% costs on firm productivity formally, a translog that a 10 percent reduction in transportation20.4 costs to cost function was estimated using the 2007 and ports would increase production by 0.79 percent.17 0% All districts Exclude Maputo/Matola 2018 Business Environment and Enterprise Perfor- < 1 hour 1-2 2-4 4-6 > 6 hours mance Survey data.16 The sample included 903 firms Figure 34. Estimated Output Elasticity with Respect to Factor Prices based in seven urban areas: Beira, Cabo Delgado, Greater Maputo, Manica, Nampula, Tete, and Zam- 0.4 Marginal effect 0.2 +1.64*S.E. bezia. The dataset is not a panel but is composed of Output elasticity -1.64*S.E. 0.0 two cross-sectional data at different points in time. It -0.2 is assumed that firm i’s production cost (c) at location -0.4 j at time t depends on input prices (W) conditional on -0.6 output (y). Four inputs are considered: labor (L), elec- -0.8 Local market Inventory Agglomeration interruption MAI Wage Power outage Input goods Transport costs Electricity tricity (E), intermediary and final goods (F), and other economies price prices Water access to port (M). The firm’s productivity may also depend on the quality of available infrastructure services (Z) and other firm characteristics (X). The quality of transport infra- Source: World Bank estimate. Note: S.E., standard error. structure is not directly measured using available data in the dataset, but it can be represented by the level of inventory that firms hold. According to the standard 66. Firm inventory is an important cost factor in pro- inventory theory, firms should have more inventory ductivity. As the conventional economic order quan- when transportation is costly and unreliable. Finally, tity model predicts, optimal firm inventory increases based on firms’ locations, transport connectivity is with the cost of purchasing goods, including transport measured in three ways: local market connectivity and handling costs (e.g., Arrow et al. 1951; Li and Li (measured according to number of people who live 2013; Shirley and Winston 2004). There is already weak within 30 minutes; LOCAL), MAI, and transport costs to correlation between firm inventory and port acces- bring goods to the nearest port (PORT): sibility: the higher transport costs to ports, the more inventory (figure 35). Moreover, holding everything ' ln +!"# = -./!" ' ' else constant, inventory is found to be costly for firms, # , *!"# , )!"# , 12341"# , 546"# , 72('"# ; :!"# ; as expected. Estimated elasticity is –0.16 to –0.17, depending on specification. If inventory increased 10 65. Although wages have the highest output elas- percent, possibly because of unreliable transporta- ticity of all inputs, port access was found to be the tion, firm output would decline by 1.6 percent to 1.7 most important determinant of firm productivity percent because more inventory costs are incurred. of infrastructure services. This is generally consistent All indications are that developing a reliable transport with the above-mentioned firm managers’ assessment network, particularly access to ports, is important for of business constraints. Output elasticity associated firms to increase their productivity. This is consistent with wages was estimated to be –0.43, meaning that a with the fact that the government of Mozambique 10 percent reduction in wages would increase output invested heavily in the primary road network during by 4.3 percent (figure 34). In the infrastructure sectors, the first half of the 2010s. 16 See a full discussion in Iimi 2021b. 17 According to the 2007 Business Environment and Enterprise Performance Survey data, approximately 17 percent of firms surveyed imported materials and equipment from abroad, and approximately 5 percent exported their products. 41 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 35. Average Firm Inventory and Port Accessibility Figure 36. Firm Distribution and Changes of number of firms 80 Average inventory (days of 70 sales or production) 60 50 40 30 20 10 0 < US$5 US$5 - 15 US$15 - 25 > US$25 Transport costs to port (US$ per ton) Source: World Bank estimate. (2011) (2018) Fever in lastIn 67. the longer 2 weeks? infrastructure term, transport If yes, visited a Fever in last 2 weeks? If yes, visited a influences firms’ decisions on where to locate, thereby fostering or suppressing No, 6% agglomeration of No, 7% businesses. No, 87% In theory, firm agglomeration or No, clustering 83% is important in increasing firm productivity (Fujita et al. 17% 13% 1999; Krugman 1991). TransportYes, 7% costs are an import- Yes, 10% ant determinant of firm investment. Despite falling eco- nomic costs of distance in recent years, many firms still prefer to be located close to each other. The evidence in the literature is generally supportive of this (e.g., Lee et al. 2012; Mare and Graham 2013; Procher 2011). In many developed countries and rapidly growing Source: INE Estatisticas Territoriais. emerging economies, firm agglomerations and indus- trial clusters have been established and have helped expand the industrial sector (e.g., Yusuf et al. 2008). experiencing some reduction in the number of firms. 68. Although major business clusters are tradition- New firms are emerging in some nontraditional areas, ally concentrated in Maputo and Beira, new firms such as Sofala, Manica, and Tete provinces (figure are emerging in northern Mozambique. Although 36).19 Although these changes are marginal compared the industrial sector, which typically benefits from with the existing business base of the traditional urban agglomeration economies, is thin in Mozambique, it is areas, they can be interpreted to mean that some important to understand where firms exist and where firms prefer not to be located in traditional areas such new ones are emerging. Public transport investment as Maputo, possibly because of their increasing con- needs to support efficient mobility. There are more gestion and high business costs. Nontraditional areas than 7,300 firms in Maputo City alone. As far as “new may be becoming more attractive, possibly because of firms” are concerned,18 the official firm registry data increasing regional integration with neighboring coun- indicate that these traditional cities seem to have been tries (see box 3 for more discussion). 18 With available data, the number of new firms is defined by a net increase 19 There is a possibility that statistical errors in the database, which is by no (or decrease) in number of firms over time. No information is available in mean perfect, cause the observed changes in the number of firms, with gross terms. many informal firms excluded. 42 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Box 3. Firm Distribution According to Type of Industry Firms are distributed differently across industries depending on their nature and require- ments. In Mozambique, the food processing sector is expected to grow. Agribusinesses, which must be located near agricultural production areas, are concentrated in Manica and Sofala provinces (figure B3.1). There are agribusiness opportunities in fruit and vegetable packag- ing, meat and dairy products, and seafood and rice processing in those provinces (Ministry of Industry and Trade n.d.). Agribusiness development is related to regional trade and integration. The Beira corridor has long been recognized as a potential regional agricultural growth corridor. Since the early 2000s, new commercial farming of horticulture and cash crops, such as sugarcane and tobacco, has been developed in Manica, Sofala, and Tete provinces, but lack of reliable trans- port infrastructure, including secondary roads and airports (e.g., Chimoio Airport, Manica and Chingozi Airport, Tete), and of agriculture-supporting infrastructure such as irrigation, storage facilities, and input supply networks continues to be a challenge in Mozambique (Beira Corri- dor 2015).20 Many firms in Manica Province export produce to Europe and South Africa through Zimbabwe. Although detailed analysis of regional trade and integration is beyond the scope of this report, it is important to examine this further. Cotton is Mozambique’s most important agricultural export crop. Current production areas include Manica, Sofala, Nampula, and Cabo Delgado (appendix B). The government of Mozam- bique is also promoting the textile industry in the northern provinces and leather in Tete Prov- ince (Ministry of Industry and Trade n.d.). Many operable and repairable mills and equipment are sitting idle. There are business opportunities in the service sector all over the country, but they tend to be concentrated in consumption areas, which are cities. The distribution of the service industry is closely associated with the distribution of the population. 20 Beira Corridor. Beira Agricultural Growth Corridor: Delivering the Potential. Available at http://www.agdevco.com/sysimages/BAGC_Investment_Blueprint.pdf 43 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure B3.1. Firm Distribution According to Industry Source: INE Estatisticas Territoriais 2014. 69. To analyze firms’ decisions on where to locate, a 70. The estimated results show that firms prefer to conventional conditional logit model is estimated (e.g., locate near other firms (agglomeration economies) McFadden 1974). The probability of firm i choosing a and where local connectivity, measured according particular location, such as district j, can be written as: to road density, is good; transport costs to ports are low; and the local market is large (table 2). The exp( X j ' b ) policy implications are straightforward: First, existing Pr( yi = j ) = å exp( X k k 'b ) clusters and cities will continue to grow, which calls for efforts to increase efficiency and sustainability in where X is a set of location-specific characteristics, urban transportation while mitigating congestion. Sec- such as local market size and labor costs. A firm is sup- ond, port connectivity is essential to keep firms com- posed to compare these attributes of districts j={1,…,J} petitive. Despite the currently limited involvement of in choosing where to locate. The sample data come Mozambican companies in international trade, there from the INE Estatisticas Territoriasis (2015), in which is some potential to explore business opportunities in 593 firms were newly registered in 75 districts from regional and global markets. To facilitate such poten- 2009 to 2012. tial, port accessibility is critical. 44 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 2. Estimated Conditional Logit Model of Firm Location Factor Coefficient of determinant (standard error) ln number of existing firms 0.451 (0.060)*** 0.137 (0.052)*** 0.091 (0.051)* Road density 0.045 (0.003)*** 0.063 (0.004)*** 0.063 (0.004)*** Transport cost to port –0.005 (0.003)* –0.014 (0.003)*** –0.016 (0.003)*** ln population –0.225 (0.084)*** 0.556 (0.093)*** 0.604 (0.092)*** ln population density –0.417 (0.048)*** –0.826 (0.076)*** –0.856 (0.074)*** ln local wage 0.300 (0.081)*** 0.305 (0.078)*** Secondary school enrollment 0.012 (0.001)*** 0.011 (0.001)*** Observations, n 43,808 43,808 43,808 Wald chi-square 426.15 643.33 676.14 Pseudo R-squared 0.066 0.095 0.099 p=*.1; **.05; ***.01. Supporting Human Capital Development 71. Health care access is limited in rural areas even 72. In Mozambique, various health indicators have in developed countries. It is often difficult for peo- improved in recent years. Life expectancy increased ple who live in remote areas to access good-quality by nearly 9 years between 2001 and 2017, from 48.8 health care services because of physical distance, lack to 57.6 years old. Infant mortality fell from 99.1 deaths of transport services, low quality of services offered, per thousand live births in 2003 to 68.1 in 2011 (World unaffordability, and low health insurance enrollment Bank 2020d). The share of children under 5 years old (e.g., Bourke et al. 2012; Gamm et al. 2003; Institute who slept under an insecticide-treated net increased of Medicine 2005). Rural residents are less likely to from 35 percent in 2011 to 73 percent in 2018 (DHS visit the emergency department than their urban 2019). The country’s health network covers the nation. counterparts (Lishner et al. 2000) and more likely to According to regional data from the World Health be hospitalized for ambulatory care sensitive condi- Organization, there are 1,579 health facilities in the tions because they do not have access to primary care country (figure 37) (Maina et al. 2019). There are three and their condition tends to deteriorate (Laditka et al. central hospitals and eight provincial hospitals. Rural 2009). Sustainable, reliable transport infrastructure is health centers and health posts account for nearly 90 thought to promote regular use of health care services percent of facilities (table 3). and prevent severe disease. 45 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 3. Health Facilities in Mozambique Figure 37. Location of Health Facilities Type Number of facilities Central hospital 3 Provincial hospital 8 District hospital 16 General hospital 5 Rural hospital 29 Urban health center A 39 Urban health center B 56 Urban health center C 49 Rural health center I 130 Rural health center II 982 Health post 262 Total 1,579 Source: Maina et al. 2019. Source: Maina et al. 2019. 73. Still, people lack access to health care services, Figure 38. Walking Time to Facilities particularly in rural areas. The health care infra- 100 2009 2015 structure is limited, medicine shortages are common Walking time (minutes) 80 (USAID 2020), and transport connectivity is limited. In 60 rural Mozambique, more than 60 percent of the pop- ulation lacks access to health care services because of 40 transportation costs and travel distance (Schiwitters et 20 al. 2015). Approximately 90 percent of Mozambique, 0 containing 67 percent of the Mozambican population, Rural Urban Rural Urban Rural Urban Rural Urban is considered unserved (Luis and Cabral 2016). In rural Transport Market Primary school Clinic Mozambique, access to socioeconomic facilities deteri- orated between 2009 and 2015 (figure 38), consistent PPP 20,000World Bank 2019a Source: with deterioration in rural accessibility discussed in the Other projects previous section. 15,000 Road safety Project preparation 10,000 Rehab/Upgrade of national roads Rehab/Upgrade of regional 5,000 roads Construct/Rehab of bridges 0 Maintenance of 46 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 39. Share of Population within 1 Hour Walking Distance of Health Care Facility (2010) (2017) Source: World Bank estimate. 74. Detailed road network data show that approx- that accessibility improved within areas close to health imately three-quarters of Mozambicans cannot facilities, whereas very remote areas were left less reach a health care facility by walking.21 There is no connected. This is consistent with the worsening of the universally accepted notion to allow people to travel quality of the rural road network in recent years. for health care. Two thresholds are examined: 1-hour walking distance and 1 -hour driving distance (Luis 75. Health care access varies significantly accord- and Cabral 2016). It is estimated that the share of the ing to region and district. Accessibility is relatively total population living within 1-hour walking distance good near urban areas and along main corridors, increased from 19.8 percent in 2010 to 24.8 percent whereas rural areas lag; accessibility is limited in in 2017 (figure 39), whereas connectivity by vehicle Niassa, Tete, Gaza, and Inhambane provinces. There declined slightly from 68.1 percent to 64.1 percent have been some minor changes in accessibility, but over the same period (figure 40). One interpretation is regional inequality seems persistent. 21 Unlike the existing literature, the quality of roads is taken into account in network analysis. 47 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 40. Share of Population within 1-Hour Driving Distance of Health Care Facility (2010) (2017) Source: World Bank estimate. 76. Malaria is a serious, life-threatening disease accessibility on health, the household response to hav- that accounts for 29 percent of all deaths and 42 ing a fever was examined. Demographic Health Survey percent of deaths in children under 5 years old in data from 2011 and 2018 were used. The former sur- Mozambique (USAID 2019). In 2018, approximately 40 veyed approximately 9,800 individuals and the latter percent of children aged 6 to 59 months tested pos- approximately 29,000 (figure 41). The probability of itive. There is major regional variation in prevalence a person having a fever increased from 13 percent in rates, from less than 1 percent in Maputo City to 57 2011 to 17 percent in 2018 (figure 42). The conditional percent in Cabo Delgado (DHS 2019). probability of a person visiting a health care facility when they had a fever increased slightly, from 53 per- 77. Approximately 17 percent of people reported cent in 2011 to 59 percent in 2018. in 2018 that they had had a fever in the previous 2 weeks, of whom slightly more than half had vis- ited a health facility. To evaluate the effect of road 80 Average inventory (days of 70 48 sales or production) INVESTING IN MOZAMBIQUE’S CONNECTIVITY: 60 TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 50 40 30 20 10 0 < US$5 US$5 - 15 US$15 - 25 >U Figure 41. Location of Households Surveyed in Transport Figure 42. Access to Health Care Services costs to by People (US$ per ton) port Fever with DHS, 2011 and 2018 80 Average inventory (days of (2011) 70 sales or production) Fever in last 2 weeks? If yes, visited a Fever in last 2 60 50 40 No, 6% 30 20 No, 87% No, 83% 13% 10 0 Yes, 7% < US$5 US$5 - 15 US$15 - 25 > US$25 Transport costs to port (US$ per ton) (2011) (2018) Fever in last 2 weeks? If yes, visited a Fever in last 2 weeks? If yes, visited a No, 6% No, 7% No, 87% No, 83% 13% 17% Yes, 7% Yes, 10% Source: DHS data. Source: World Bank estimates based on Demographic Health Survey data. 78. The probability of a person vising a hospital is 79. People who lived closer to a health care facility high near urban areas and along main corridors but were twice as likely to visit a health care facility to less than 10 percent in certain rural areas. Health seek advice or treatment for fever. The likelihood accessibility varies significantly according to location increased with accessibility to health facility (figure 44). even though the samples were not designed to be If a health care facility was less than 20 minutes away representative at the district level (figure 43). Accessibil- 70 percent of people would seek advice there, but if ity was particularly low in Niassa, Tete, and Gaza prov- it was more than 2 hours away, only 40 percent to 50 inces, which is broadly consistent with the geographic percent would do so. Malaria accounts for 30 percent distribution of rural connectivity discussed above. of total deaths in Mozambique, indicating the impor- tance of road connectivity in saving lives. 49 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 43. Share of People Who Had a Fever and Visited a Health Care Facility (2011) (2018) 0.8 0.70 Share of those who visited 0.7 0.65 0.61 health facility (0 to 1) 0.6 0.5 0.47 0.41 0.4 0.3 0.2 Source: World Bank estimates based on DHS data. 0.1 0.0 < 5 min 5 - 20 20 - 55 55 - 160 > 160 min Figure 44. Share of People who Visited a Health Care Facility According to Distance Estimated walking time to the nearest health facility (2011) (2018) 0.8 0.70 0.8 0.68 0.67 Share of those who visited Share of those who visited 0.7 0.65 0.7 0.62 0.61 health facility (0 to 1) health facility (0 to 1) 0.6 0.6 0.55 0.53 0.5 0.47 0.41 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.1 0.1 0.0 0.0 < 5 min 5 - 20 20 - 55 55 - 160 > 160 min < 5 min 5 - 20 20 - 70 70 - 140 > 140 min Estimated walking time to the nearest health facility Estimated walking time to the nearest health facility Source: World Bank estimates based on DHS data. 0.8 0.68 0.67 of those who visited 0.7 0.62 lth facility (0 to 1) 0.6 0.55 0.53 0.5 0.4 0.3 0.2 50 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 45. Estimated Marginal Effects of Factors on Likelihood of Visiting a Health Care Facility 0.15 Marginal effect Marginal effect +1.64*S.E. 0.10 -1.64*S.E. 0.05 0.00 -0.05 e e cle 18 d d h h r ot e Ca al Ag 2n 3r 4t 5t / M tim 20 cy m Fe t= or el av Tr Wealth index percentile ke Bi Source: World Bank estimates. 80. More formally, a probability regression model 81. A 10 percent reduction in accessibility to a was applied to estimate the net effect of greater health care facility could increase the likelihood transport connectivity on decisions about whether of a person visiting one by 0.2 percentage points. to visit a hospital, controlling for other household The marginal effect of transport accessibility was esti- and individual characteristics.22 In our data, there was mated at 0.029 in absolute terms. Other factors were wide variation in transport connectivity according to also important in determining accessibility to health location and time. This variation constitutes the basis care services, with household income being the largest of empirical identification, although the data are not determinant. The richest households have the great- a panel but cross-sectional data at two points in time. est accessibility. Younger people were more likely than Transport connectivity (Z), characteristics of the near- older people to visit a health care facility if they had est health facility (V), and individual and household a fever. Transport means of ownership is also instru- characteristics (X) determine the probability of individ- mental. People were more likely to visit a hospital if ual i seeking advice at health care facility j at time t. A their household owned a bicycle or motorcycle. House- probit mode was used to estimate this:23 hold sanitary and dwelling conditions also matter. Over the survey period, general health care accessibil- ity seems to have improved: The coefficient of time t ∗ 1 E- C!"# ' = *!# F + )!# ' %+ + G ' "# %, + +# + ,!"# > 0 C!"# = D was significantly positive (figure 45). 0 JKℎMNOEPM 22 See a full discussion at Iimi (2021c). 23 The conventional exogeneity test indicated that our transport connectivity variables were largely exogenous. 51 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 82. To increase health care accessibility, it is Figure 46. Potential Priority Roads for Health Care Access important to increase rural connectivity. The major- ity of roads that connect communities to local health care facilities are unpaved roads. United Nations Office for the Coordination of Humanitarian Affairs data iden- tify 24,602 settlements in Mozambique.24 The average road distance from a settlement to the nearest health care facility is approximately 10 km. Approximately 8,800 settlements are more than a 1-hour drive away from the nearest health care facility. There are specific roads that are needed to be rehabilitated for local peo- ple to be connected to their nearest health facilities. To connect these settlement areas better, approximately 5,500 km of roads must be improved, of which 5,400 km are unpaved and 2,800 km are in poor condition (figure 46). These roads can be considered a priority from the health care access point of view. They do not have to be paved but must be passable for bicycles and motorcycles. Niassa, Tete, Gaza, and Inhambane provinces are among the top priority areas. Source: World Bank estimates. 24 Information from the Office for the Coordination of Humanitarian Affairs database (accessed in October 2020) https://data.humdata.org/dataset/mozambique-settlement-shapefiles 52 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Building More Resilience in Road Network 83. Mozambique is highly vulnerable to extreme scenario, climate risk exposure of the road network climate events, and the poorly maintained road net- would not be much different than under the current work increases this vulnerability. Cyclones Idai and climate scenario, but under a high climate change Kenneth severely damaged 3,600 km of roads, caus- scenario, it is likely that exposure would be substan- ing significant disruptions in the transport system and tially greater. Twenty-year floods would affect approx- the entire economy. The government would have to imately 14.3 percent of roads, and 100-year floods expend resources to develop and sustain the network would affect more than 20 percent of roads. and add resilience. Given the limited fiscal space, sus- tainability is critical to maximizing economic benefits from better connectivity while minimizing costs. Figure 47. Flood-Prone Areas (25-Year Flood) 84. The road network is of particular importance in avoiding critical disruptions in access to socioeco- nomic services. As discussed above, people must be within 20 minutes travel time of a health care facility when it is needed. Some roads are more important than others because there are no alternative routes. This is often the case in remote rural areas. Travel time costs are greater with variability and uncertainty of arrival time (e.g., Cohan and Southworth 1999). 85. Mozambique’s road network is exposed to cli- mate risk. Under current climate conditions, Gaza, Sofala, Zambezia, and Nampula are particularly vulner- able to heavy rainfall and floods (figure 47). According to a projection with a 20-year flood recurrence interval, approximately 11 percent of roads are located in flood- prone areas (table 4), and approximately 30 percent are in poor condition and consequently are considered to be particularly vulnerable. Although available cli- mate models have much uncertainty, according to the World Ban Climate and Disaster Risk Screening Tools, all indications are that the temperature will increase in Mozambique and that there will be more rainfall Sources: ANE; GAR 2015 Atlas Risk Data and Software Download Facility. during the wet season. Under a low climate change 53 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 4. Length of Roads in Flood-Prone Areas Road Existing Current climate Climate change scenario condition road network Low Medium High Recurrence interval (years) 5 20 100 5 20 100 5 20 100 5 20 100 Kilometers Paved 8,138 452 672 955 433 647 931 486 723 1,018 543 800 1,118 Unpaved 21,656 1,665 2,836 4,215 1,564 2,699 4,075 1,868 3,097 4,531 2,204 3,475 4,959 Total 29,794 2,117 3,508 5,170 1,997 3,346 5,007 2,354 3,820 5,549 2,746 4,274 6,077 Infrastructure risk is determined according to standards. Poorly maintained unpaved roads tend to exposure to climate events and vulnerability of be more vulnerable, whereas road structures such infrastructure. (For more detail, see Espinet and as bridges and culverts can reduce vulnerability. The Rozenberg 2018.) A road is exposed to flood risk economic risk of extreme climate events is determined if it is located in a flood-prone area. The degree of according to infrastructure risk and degree of eco- exposure can be assessed based on flood maps with nomic activities that disrupted infrastructure services various recurrence intervals. Vulnerability is deter- may affect (economic exposure). mined according to technical road specifications and 6Q-NRPKNS+KSNM NEPT = 3UEVRKM MWXJPSNM × GSUQMNRZEUEK: [+JQJVE+ NEPT J- MWKNMVM M#MQKP = 6Q-NRPKNS+KSNM NEPT × [+JQJVE+ MWXJPSNM 86. The economic risk of flood events is particularly SPAM. Agriculture is an important sector in Mozam- high in Zambezia, Nampula, and Inhambane. For bique’s economy and one of the most affected sectors. simplicity, only a 25-year flood scenario is considered Each parameter is normalized to a range of 0 to 1, and under the assumption that roads in poor condition then the three indices are multiplied (Rozenberg et al. are vulnerable. For each district, climate exposure is 2017; Espinet et al. 2018). Economic risk is generally measured according to the ratio of roads located in high in Zambezia, Nampula, and Inhambane Province. flood-prone areas. Vulnerability is measured according Additional efforts are required to build resilience in the to ratio of roads in poor condition. For economic expo- transport infrastructure (figure 48). sure, the total value of crop production is used from 54 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 48. Economic Risk of Extreme Climate Events According to District • Access to domestic markets is an important con- straint on agricultural production. There are several priority areas in Inhambane, Niassa, Tete, and Man- ica where there is untapped agricultural potential but market accessibility is low. • In addition to local market connectivity, access to ports is essential for firms to increase their produc- tivity and foster agglomeration economies, which creates formal jobs in the local market. This sug- gests that improving and maintaining primary cor- ridors between major cities as well as regional hub ports is important. Inventory costs affect productiv- ity, and a poorly maintained, unreliable transport system increases firms’ inventory, reducing their productivity. • Accessibility to a health care facility is an important determinant of whether people seek health care services, which could be of considerable conse- quence for human capital development in Mozam- bique. There are specific local needs to connect people to health care facilities. Approximately 5,500 km of feeder roads must be improved to ensure Source: World Bank estimate. that everyone has access to a health care facility within 1 hour’s drive. • The economic risk of extreme climate events such as floods is particularly high in Zambezia, Nampula, Section Summary and Inhambane provinces; additional efforts are required there to increase resilience in the trans- port infrastructure. 87. Key takeaways from this section are as follows. • Strategic prioritization based on evidence is essen- • Different types of roads are required for different tial to maximize the effects of transport connectiv- development objectives, such as economic growth, ity, especially under a budget-constrained environ- rural development, and poverty reduction. The road ment. From the empirical point of view, each of the network should be developed as a seamless net- effects examined in this section is a partial equilib- work from tertiary to primary roads. rium and cannot be overgeneralized to examine the overall effects or priority. It is important to develop • For any development objective, transport infrastruc- a strategic prioritization framework consistent with ture is one of many constraints. To stimulate eco- the government’s national development strategy. nomic growth, coordinated efforts are required to (See section V for further discussion.) address all of them simultaneously. 55 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 4.  Current Road Spending 56 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Institutional Set-Up for Public Expenditure on Roads 88. In Mozambique, the Road Fund plays an import- 89. Unlike in many other Sub-Saharan African ant role as the central institution in the road asset countries, since becoming an independent institu- management system (figure 49). The Road Fund tion, the Road Fund has consolidated internal and was created in 1989 as a unit of the Ministry of Public external funding sources to finance road mainte- Works, Housing and Water Resources. In 1999, ANE nance and development. Its internal resources come was created as an implementing agency of road works. mainly from Treasury subsidies (annual appropria- In 2003, the Road Fund became an independent institu- tion from the national budget), fuel levies, and other tion of the Ministry of Public Works, Housing and Water road user charges. External resources include grants Resources, with its own legal, administrative, financial, and credits from bi- and multilateral donors that are and patrimonial autonomy. Two ministries oversee normally earmarked for the road sector. From 2010 to the Road Fund: Ministry of Public Works, Housing and 2018, an average of nearly 40 percent of road sector Water Resources for technical oversight and Ministry resources were internal, which is relatively high by of Economy and Finance for financial oversight (Fundo African standards.26 de Estradas 2018).25 The Road Fund is responsible for official road networks and unclassified roads. Figure 49. Overview of Institutional Set-Up of Road System 25 Decreto no 61/2019. Ajusta as atribuçaões, gestão, regime orçamental,tutelar, organizacão e funcionamiento de Fundo de Estradas, criado pelo decreto no 22/2003. 26 For instance, although available data are fragmented, external resources for road investment account to 55 percent to 90 percent of total capital expendi- ture in Africa (Gwilliam 2011). 57 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 90. ANE is the major implementing agency for road Development Goal Target 9.1. Even achieving a min- projects, consuming on average 96 percent of total imum standard, such as gravel road, costs approxi- road sector resources from 2010 to 2018. Two enti- mately USD79,000 per km in Mozambique, for a total ties—ANE and local governments—implement road of USD10.2 billion. Substantial resources are also programs in Mozambique. The former is responsible needed to keep the current road network well main- for the official road network and the latter for unclas- tained—up to USD150 million per year. (See details in sified roads. ANE is a public institution with legal and the following sections.) administrative autonomy overseen by the Ministry of Public Works, Housing and Water Resources that 92. Before the 2016 crisis, the government devoted implements national road programs, including main- significant resources to the road sector, although tenance (routine and periodic), rehabilitation, and (re) donors supported the majority of this effort. The construction of classified roads. ANE is also the line government has dedicated on average 19 billion met- agency responsible for implementing road safety icals (USD500 million; 3.6 percent of GDP) annually to initiatives and the contracting authority for national implement the national road program and transfer it road public-private partnerships (PPPs). Most of ANE’s to local governments from 2010 to 2018 (figure 50). resources come from Road Fund transfers.27 ANE Mozambique has consolidated internal and external issues technical guidelines and provides technical sup- funding sources under the Road Fund, whereas the port for local governments in implementing rehabilita- central governments of many other African countries tion and maintenance of unclassified roads. still hold the external resources. Approximately 60 percent of Road Fund resources come from the inter- national donor community, leaving approximately USD200 million (1.4 percent of GDP) to come from How Much Is Needed? internal resources. This is probably among the highest How Much Is Available? in Sub-Saharan Africa. There are no systematic data, although, on average, Sub-Saharan African countries spend approximately 1.0 percent of GDP on roads. 91. Substantial financial resources are needed to Even for resource-rich countries such as Mozambique, close the gaps in transport connectivity—as much an estimated 1.3 percent of GDP is spent on roads (Bri- as USD10 billion. Given the condition of the current cenio-Garmendia et al. 2008). More recent data show road network, it is estimated that USD940 million that resources available for roads in Sub-Saharan Afri- would be needed to rehabilitate the existing road can countries account for on average 0.4 percent of network (without any upgrading). For universal access GDP (figure 51). Although the scope of the road fund purposes, another USD10 billion would be needed to mandate may differ between countries, this is also reclassify and construct currently unclassified roads. a good proxy for road expenditures and shows that Given current population density, approximately Mozambique spent the largest share of GDP on roads 129,000 km of “new” roads would be needed to sup- of African countries. port the universal access envisaged in Sustainable 27 Government regulations enable ANE to generate other resources, for example, fees from advisory services to support local governments for managing the road network under their purview (see article 27 of Decree No. 12/2007). Although no detailed data on such additional resources were available, they are presumably much less than transfers from the Road Fund. 58 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 93. After the 2016 debt crisis, the road sector budget was sharply decreased and remained low for the second half of the 2010s. Available internal resources diminished as the economy slowed with the exchange rate sharply depreciated and financial mar- 25,000 62.1 70 kets in difficulty. Donor 62.0 57.8 contributions, which account Resources collected (MT million) Share of internal resources (%) 50.5 50.9 60 for20,000 48.6 of the total road 60 percent 45.2 budget, seemed to 50 stable, albeit lumpy as usual, although internal be 15,000 35.0 40 resources fell substantially from USD350 million (2.1 30 10,000 percent of GDP) in 2014 to USD90 million (0.8 percent 20 5,000 of GDP) in 2018. As a result, the government faced 10 a significant 0 challenge in developing and maintaining 0 2012 even 2010 2011 with the road network 2013 2014 fewer2015 2016 2017 2018 resources. External resources Internal resources Share (%) Figure 50. Overall Road Sector Budget (Planned) 35,000 Road fund budget Internal resources 5.0% (% of GDP) (% of GDP) 4.5% Road fund budget (MT million) 30,000 4.0% 25,000 3.5% 20,000 3.0% 2.5% 15,000 2.0% 10,000 1.5% 1.0% 5,000 0.5% 0 0.0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 Source: ANE. Note: GDP, gross domestic product. Figure 51. Road Fund Resources in Selected African Countries, 2017 2.5 RF resources per GDP (%) External resources 2.0 Internal resources for Mozambique 1.5 1.0 0.5 0.0 S e al i me n a Eth n da nz i Ch o e Ke ia ia Bu d’lv go Gu on Ni a i a, ar M a-Bi di Le ia M ny a S ie Zam nia Na iqu e o, n na re Ga d aga sau Rw The ine ru . te To l ad Co Be er ga Ta al aw Gu Bu Rep ine s on ng n i b i op rr a b n mb s c rk i oi g Fa M Ca h a ro a mi ne a s a mb G a oz Cô M Source: World Bank 2019b. 100% 90% 80% 70% 60% 50% 40% 30% 20% 59 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Where Does the Need Exist? Where Does the Government Spend? 94. To ensure better road access to all Mozam- needs are higher where connectivity is lower (figure bicans, enormous resources are needed where 52). The need is highest in Niassa, followed by Tete connectivity is low, for example, Niassa, Nampula, and Zambezia (figure 53), because these provinces Tete, and Zambezia. To achieve universal access, it have low population density, low road density, and is estimated that these four provinces would require poor-quality roads. As a result, more resources are more than half of road investment needs (table 5). The required to expand and improve the network. Table 5. Estimated Financial Need for Universal Road Access Province Length of roads (km) Climate change scenario Share (%) Existing Needs to be Rehabilitation Reconstruction Total classified and of existing reconstructed roads Cabo Del-gado 2,824 8,872 122 701 823 7.4 Gaza 2,555 9,447 34 746 780 7.0 Inhambane 2,860 10,285 67 813 880 7.9 Manica 2,515 11,058 35 874 909 8.1 Maputo 1,697 3,098 69 245 314 2.8 Nampula 4,103 13,203 144 1,043 1,187 10.6 Niassa 3,650 26,622 165 2,103 2,268 20.3 Sofala 2,459 12,729 33 1,006 1,039 9.3 Tete 2,931 17,068 80 1,348 1,428 12.8 Zambezia 4,236 16,949 190 1,339 1,529 13.7 Total 29,831 129,332 940 10,217 11,157 100.0 Source: World Bank estimates. 60 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Chart Title 30.0 20.0 10.0 95. The current distribution of public resources because of misallocation of public infrastructure 0.0 is not necessarily -10.0 -5.0 0.0 consistent 5.0 with 10.0 the underlying 15.0 20.0 25.0 investments between urban and rural areas (World -10.0 need for road connectivity. Inequality has increased Bank 2019a). Not surprisingly, the urban-rural divide in in Mozambique-20.0 (e.g., Gradin and Tarp 2019), mainly transport increased between 2010 and 2017. Although -30.0 primary paved roads were well maintained, unpaved Figure 52. Estimated Financial Need and Rural Accessibility roads deteriorated, and rural accessibility declined (fig- 400 ure 54). The Economic and Social Plan/Integrated Road universal access (US$ million) 350 Sector Program annual reports do not provide detailed Estimated financial need for 300 expenditure data at the provincial level but report 250 progress of maintenance projects in terms of length of 200 150 roads for each province. Zambezia, Inhambane, Man- 100 ica, and Sofala implemented more routine road main- 50 tenance projects in terms of their road network size 0 0 10 20 30 40 50 60 (figure 55) than Nampula, Niassa, and Tete, where it is Rural Access Index (%) estimated that there is more unmet need. Source: World Bank estimates. Figure 53. Estimated Need According to District Figure 54. Urban-Rural Divide in Transport Accessibility 50 50.0 48.7 45 40 35 30 24.6 25 20.4 19.3 21.6 20 15 8.4 10 5.9 4.8 5.2 5 0 2010 2017 2010 2017 2010 2017 2010 2017 2010 2017 Paved road in Unpaved road in TR cost to market RAI (%) TR cost to port good condition good condition ($/ton) ($/ton) (%) (%) Source: World Bank estimates. 35 34 Mortality caused by road traffic Mozambique injury (per 100,000 people) Figure 33 55. Maintenance Projects and Rehabilitation Need 32 According to Province Tanzania 31 30 Madagascar 29 maintenance Routine in 282018 (km) Kenya 27 26Existing roads Sub-Saharan 25 (km) Africa 2000 2005 2010 2015 Financial needs ($ mil) Rehab/ Reconst needs (km) 0% 20% 40% 60% 80% 100% Zambezia Nampula Niassa Tete Inhambane Cabo Delgado Gaza Manica Source: World Bank estimates. Sofala Maputo Actual 61 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 56. Routine and Periodic Maintenance Projects According to Province Routine Maintenance - paved roads Routine Maintenance - unpaved roads 1200 4000 1000 3500 3000 800 2500 km 600 2000 km 400 1500 200 1000 0 500 0 za ne o a e la o la sa z ia ic ad t ut pu fa Ga as Te ba an bé ap a e o la z ia lg za ne o So la sa Ni m ic t m ut ad fa M De pu m Te Ga as M ba an bé Na ap ha So Za lg Ni m m bo M m De In M Na ha Za Ca bo In Ca 2012 2013 2014 2015 2016 2017 2018 2012 2013 2014 2015 2016 2017 2018 Periodic Maintenance - paved roads Periodic Maintenance - unpaved roads 100 200 90 180 80 160 70 140 60 120 km 50 km 100 40 80 30 60 20 40 10 20 0 0 a e o la z ia za ne o la sa za ne o a e la o la sa z ia ic t ut ad ic ad fa t ut pu Te pu fa Ga as Ga as ba an Te bé ba an bé ap So lg ap lg So Ni m Ni m m M m m De M De M m M Na Na ha Za ha Za bo bo In In Ca Ca 2012 2013 2014 2015 2016 2017 2018 2012 2013 2014 2015 2016 2017 2018 96. In some provinces, routine maintenance on mismatch between needs and resources. To pro- unpaved roads has been reduced or stagnant, mote decentralization of road management, one of which raises concerns about sustainability of the the fundamental principles of the Road Sector Policy road network. There are some variations in peri- in Mozambique (Resolucão no 61/2008), the Road odic and routine maintenance progress according to Fund dedicates approximately 4 percent of its internal province over time (figure 56). Regional differences resources to co-financing road programs of district in levels of maintenance activities remain broadly and municipal road authorities. Approximately half of unchanged. Zambezia, Inhambane, and Sofala main- this is evenly distributed to 154 district governments tained more roads than other provinces, but routine and used for maintenance and rehabilitation of unclas- maintenance on unpaved roads in Zambezia declined sified roads that districts administer and purchase of substantially, as it did in Tete, Nampula, Cabo Delgado, materials and equipment necessary for maintenance and Niassa. Lack of road maintenance is likely to accel- and rehabilitation. In addition, by law, 10 percent of erate deterioration of the road network and increase Road Fund revenue stemming from the fuel levy is its vulnerability to climate events. transferred to municipalities to finance rehabilita- tion of urban roads. Although the current transfer 97. The resource allocation mechanism could be mechanism is transparent, the method of allocating improved by accounting for actual needs. For resources may need to be based more on need, for decentralization purposes, the government transfers which a more comprehensive road inventory database some road sector resources evenly to local govern- including classified and unclassified roads must be ments, which make sense from an equity perspective developed and updated regularly. but may be causing the above-mentioned geographic 62 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING What Is the Government Spending? 98. In the late 2000s and early 2010s, the govern- 64 percent in 2014 (figure 57). Just before the debt cri- ment of Mozambique spent a substantial amount sis in 2016, some of these investments were replaced of money on rehabilitating and upgrading national with PPP projects, which normally include rehabilita- roads. Rehabilitation consists of many different types tion projects as the private sector’s initial contribu- of projects, such as realignment, work on the subbase, tion. Together with these PPP projects, rehabilitation and pavement replacement. Upgrading mostly con- and upgrading of national roads accounts for the vast sists of paving previously unpaved roads. These types majority of the total budget executed. External funding of activities account for on average USD174 million per sources cover approximately 60 percent to 70 percent year (table 6), accounting for approximately 42 percent of these expenditures. of the executed total road fund budget, with a peak of Table 6. Estimated Financial Need for Universal Road Access Activity Average 2010 2011 2012 2013 2014 2015 2016 2017 2018 2010-18 2016-18 Administration 20.3 18.0 19.1 24.8 23.7 18.5 14.7 14.8 17.5 19.0 15.7 Capacity building 4.6 6.7 6.2 3.8 2.7 2.7 2.9 2.7 0.8 3.7 2.1 Maintenance of roads and 119.9 93.4 112.0 121.5 121.3 122.1 58.4 60.1 128.2 104.1 82.3 bridges Construction and rehabilita- 12.7 104.6 15.0 41.3 24.5 23.4 18.4 3.2 5.0 27.6 8.9 tion of bridges Rehabilitation and upgrade 21.8 23.4 23.9 14.4 21.4 12.9 6.1 9.1 5.5 15.4 6.9 of regional roads Rehabilitation and upgrade 118.6 179.8 219.8 302.3 433.9 126.2 37.7 61.4 86.4 174.0 61.8 of national roads Project preparation 0.0 0.0 0.9 0.2 0.4 0.2 0.1 1.4 15.7 2.1 5.7 Road safety 2.7 2.1 6.7 2.1 2.0 1.2 0.3 0.3 0.4 2.0 0.4 Other projects 0.0 0.0 0.0 0.0 43.9 24.3 3.9 5.0 1.9 8.8 3.6 Public-private partnerships 0.0 0.0 0.0 0.0 0.0 105.3 27.5 25.8 9.7 18.7 21.0 Total 300.7 427.9 403.4 510.3 673.8 436.7 170.0 183.9 271.1 375.3 208.3 % of gross domestic product 2.7 3.0 2.5 3.0 3.8 2.7 1.4 1.4 1.8 2.5 1.6 Source: ANE. PPP 63 20,000 Other projects 100 TOWARD 2009 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: EFFICIENT, SUSTAINABLE ROAD SPENDING 2015 Road safety Walking time (minutes) 15,000 80 Project preparation 60 10,000 Rehab/Upgrade of national roads 40 Rehab/Upgrade of regional 5,000 roads 20 Construct/Rehab of bridges 0 0 Maintenance of Rural Urban Rural Urban Rural Urban Rural Urban Figure 57. Share roads/bridges Figure 58. Expenditures According to Activity (million 2010 2011 of Expenditures 2012 According 2013 2014 2015 to 2016 2017 Activity 2018 (% of Capacity building Transport Market Primary school Clinic meticals) executed budget) PPP 100% 20,000 Other projects 39.4 42.0 31.9 54.5 59.2 22.2 33.4 64.4 Road safety 50% 28.9 15,000 Project preparation 39.9 34.4 32.7 47.3 21.8 27.8 23.8 28.0 10,000 Rehab/Upgrade of national 18.0 roads 0% Rehab/Upgrade of regional 2010 2011 2012 2013 2014 2015 2016 2017 2018 5,000 roads Construct/Rehab of bridges Administration Capacity building Maintenance of 0 roads/bridges Maintenance of roads/bridges Construct/Rehab of bridges 2010 2011 2012 2013 2014 2015 2016 2017 2018 Capacity building Rehab/Upgrade of regional roads Rehab/Upgrade of national roads Project preparation Road safety Other projects PPP 99.  Despite the reduction in available resources devastating cyclones hit the Mozambican economy, since the crisis, the government has prioritized and the COVID crisis and the global economic down- road and bridge maintenance, which includes trans- turn are intensifying the uncertainty of the future. fers to districts and municipalities, routine and peri- Available resources for the road sector are likely to stay odic maintenance of paved and unpaved roads and flat and possibly decrease. Because of recent extensive bridges, emergency projects (e.g., to repair damage capital investment in the road network, maintenance caused by weather), and spot improvements and is a needs will increase in the coming decades. good way to preserve road assets. Before the crisis, it accounted for on average 20 percent to 30 percent of 101.  The road safety budget is far below the inter- the total budget executed. After the debt crisis and the national norm and has been declining in recent economic slowdown, available resources for the road years. This includes activities such as maintenance and sector were nearly halved. Still, maintenance spending replacement of road signs, maintenance and installa- was kept relatively unchanged (figure 58), and its share tion of axle load control infrastructures, and soft activ- increased to 47 percent in 2018. Internal sources of ities such as road safety campaigns and staff training. funding cover approximately 80 percent of this, indi- It represents on average approximately 0.5 percent of cating that the government is strongly committed to the executed total budget and is almost entirely cov- preserving existing road assets. ered by internal resources. Although the government spent approximately 100 million meticals per year on  The government of Mozambique continues to 100. road safety from 2010 to 2014, annual expenditures face the challenge that the need for road mainte- declined to approximately 20 million meticals from nance is likely to increase even more. As discussed 2016 to 2018, which raises concern because road traf- above, the road sector budget, and thus expenditures, fic deaths in Mozabmique are among the highest in declined substantially after the crisis. In addition, the the region, although they are decreasing (figure 59). 40 35 30 24.6 20.4 64 25 20 19.3 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 15 8.4 10 5.9 4.8 5.2 5 0 2010 2017 2010 2017 2010 2017 2010 2017 2010 Paved road in Unpaved road in TR cost to market RAI (%) TR cos good condition good condition ($/ton) ($ (%) (%)  The African Road Safety Action Plan 2011–2020 102. Figure 59. Road Traffic Deaths in Selected Countries that the heads of state of all African countries 35 adopted in 2012 stipulates that they allocate 10 34 Mortality caused by road traffic Mozambique injury (per 100,000 people) percent of road investment and 5 percent of road 33 maintenance resources to road safety (Small and 32 Tanzania 31 Runji 2014). If Mozambique had done this, it would 30 Madagascar have allocated approximately 520 million meticals 29 (USD8-9 million) per year for road safety, far more than 28 Kenya the government is currently spending. Road safety 27 should not be compromised. 26 Sub-Saharan 25 Africa 2000 2005 2010 2015 Sources: WDI archive for 2000 and 2005; WHO (2018) for 2013 and 2016. Section Summary 103.  Key takeaways from this section are as follows. (1.5 percent of GDP), in recent years, presenting the government with new challenges. With far fewer • Mozambique has a solid road financing structure. resources, the road network, extended by recent Although the Road Fund manages internal (road capital investment, must be maintained and devel- user charges, national budget appropriation) and oped further, including primary and rural roads. external (external loans, donor grants) resources in the road sector, ANE is responsible for imple- • Unmet need for road connectivity remains enor- menting the vast majority of road programs in the mous in Mozambique. Approximately USD11 billion country. would be needed to rehabilitate and reconstruct roads to achieve universal access. Another USD400 he government devoted substantial resources to • T million would be needed to maintain the current the road sector in the late 2000s and early 2010s, road network optimally. averaging USD500 million (3.6 percent of GDP) per year, among the highest in Sub-Saharan Africa, • There is increasing concern about underspending but external resources still fund approximately 60 on maintenance of the road network. percent. • Because of the current uniform resource redistri- uring the 2010s, the government spent a sub- • D bution to local governments, there is some geo- stantial amount of money on rehabilitating and graphic mismatch between allocated resources and upgrading national roads, with a peak of 64 percent needs. There may be potential to improve the prior- in 2014. This has contributed to the recent substan- itization and budget allocation mechanisms. tial increase in accessibility of markets and ports but has also generated additional need for road • The government’s allocation to road safety is far maintenance. below the international norm (5-10 percent of road investment and maintenance) and has been declin- • After the debt crisis in 2016, the road sector budget ing in recent years, raising serious concerns about was nearly halved, to approximately USD200 million road safety in Mozambique. 65 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 5.  Toward More Sustainable Road Financing 66 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Further Strategic Prioritization 104.  To enhance the sustainability of road sector road network. Finally, road sector revenues should be financing, seven issues in the following four areas can increased. Potential sources range from internal road be considered (figure 60). First, it is important to avoid user charges to private sector contributions. unnecessary expenditures and focus on areas where socioeconomic benefits are substantial. This can also 105.  Given potentially competing development be considered part of road asset management but objectives and limited public resources, it is essen- should go beyond that. Multisectoral strategic priori- tial to prioritize road investment (and maintenance) tization at the national level is critical, involving other from a strategic point of view. The economic analy- economic and social sectors (e.g., agriculture, health, ses in section III, although not exhaustive, have high- education, private sector development). Second, it is lighted potential priority areas : Cabo Delgado, Niassa, important to use available resources wisely; there are Tete, Zambazia, and perhaps Gaza (figure 61), although several ways to increase efficiency in the government’s any new investments (road rehabilitation and upgrad- administrative systems. Third, within the road sector, ing projects) will increase the financial need for road the government should allocate resources to the right maintenance in the future. As shown below, the cur- activities. By implementing a road asset management rent paved road network is already undermaintained. system, road sector expenditures could be optimally Public resources dedicated to the road sector have also allocated, including to resilient investment in the shrunk. Therefore, strategic prioritization is critical. Figure 60. Areas of Sustainability of Road Sector Financing Expenditures • Strategic prioritization of investment • Efficiency in budget execution Administration • Efficiency in public procurement • Optimization of road expenditures Road asset management • Resilience in road infrastructure • Increase and diversification of revenues Fiscal sustainability • Mobilization of private resources 67 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 61. Selected Indicators of Connectivity, Economic Potential, and Vulnerability Rural access index Market Access Index Health access Economic climate risk Agricultural potential Firm location Health facility Flood prone areas Source: World Bank estimate. 106.  The government has developed a road plan- agricultural values, which are derived using the Inter- ning tool to prioritize its road investments. It is national Food Policy Research Institute SPAM. The essentially a multicriteria analysis framework based Ministry of Agriculture provided crop production data on seven criteria: network criticality, traffic volume, for 2015/16. Fishery production, which is less relevant poverty, proximity to potential agricultural clusters, than crop production in the current case, is based on current agricultural production, fishery production, a global database (Halpern et al. 2015). Flood risk is and flood risk. It is important to keep updating and measured as expected annual losses according to the fine-tuning the framework as the economy grows. Pov- probability distribution of flood events with 10 inten- erty is measured according to the weighted average of sities (5-, 10-, 20-, 50-, 75-, 100-, 200-, 250-, 500-, 1,000- poverty headcount ratios when a road extends across year return period) and for four climate scenarios more than one district. Agricultural potential is mea- (one based on past climate and three climate change sured according to the weighted average of potential projections). 68 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 107.  The prioritization tool must be applied more strategically to make road investments even more selective because of the continued tight fiscal space, growing uncertainty in the global economy, and unforeseen external shocks. There is much uncertainty related to COVID-19 and climate-related change, and the outlook for global commodity markets is unclear. As new data and technologies become available (box 4), the framework should be kept updated, and adjusted if needed, to use limited available resources effectively to achieve long-term inclusive growth. Box 4. New Road Sector Digital Technologies New digital technologies could help in collecting necessary data in a timelier manner. In the road sector, the Administração Nacional de Estradas has used some of the new technol- ogies, such as mobile phone–based applications, to update the road condition database (e.g., RoadLab; figure B4.1) and assess road safety. High-resolution satellite imagery (figure B4.2) and call detail record data can potentially be used for road condition and mobility assessment (e.g., World Bank 2016). Digital development can also benefit the economy along agricultural value chains, complemented by transport and logistic infrastructure development. Figure B4.1. RoadLab Application Figure B4.2. Road Condition Assessment Using Satellite Imagery Source: World Bank 2016. 69 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Increasing Efficiency in Budget Execution 108.  The government must increase efficiency in 110.  The government should also be focused on budget execution. The execution rate of the road sec- building capacity to manage complex road and tor budget, which is generally relatively high, has been bridge investment projects, even if external donors on a steep downward trend (figure 62). Before 2016, finance them. For various reasons, such as technical the execution rate was greater than 80 percent, which complexity and cumbersome institutional arrange- is mostly attributable to timely disbursements from ments, execution rates for road and bridge rehabilita- donors to cover capital expenditures on bridge reha- tion and construction projects tend to be low (figure bilitation, but the execution rate dropped to 40 percent 63). Inefficient budget execution places significant to 60 percent after the debt crisis. The recent under- fiscal costs on the country. Better, more realistic proj- performance is mostly because of the lower-than-ex- ect planning is needed, with implementation capacity pected execution of externally funded projects but also increased further. Medium-term budgetary arrange- decreasing internal resources. ments may be one way of keeping flexibility to accom- modate unexpected events and ensuring efficient use 109.  A more solid and systematic project of limited resources while increasing the credibility of assessment and monitoring system must be built, budget execution. This is particularly useful in coun- following a high-level macro-fiscal management tries where external conditions are unstable and con- improvement initiative. In general, major road con- trol over in-year expenditures is insufficient (IMF 2013). struction, upgrading, and rehabilitation projects are mostly financed with external resources and screened Figure 62. Total Budget Execution Rate in stringent economic, environmental, and social 35,000 Total budget 140% assessments, although as the sudden discovery of 30,000 Actual spending 120% Budget and actual spending (MT Execution rate (%) nonperforming external public loans in 2016 revealed, Budget execution rate (%) 25,000 100% the government’s project assessment and governance 20,000 80% million) systems could stand to be strengthened. With the 15,000 60% assistance of the donor community, the government 10,000 40% embarked upon a macro-fiscal management improve- ment initiative, Gestao Economica para Desenolvi- 5,000 20% mento Inclusivo, in 2019, with the objectives of devel- 0 0% 2010 2011 2012 2013 2014 2015 2016 2017 2018 oping a medium-term fiscal framework for inclusive growth, improving the public investment management Figure 63. Budget Execution Rates According to Activity system by standardizing public investment appraisal practices, and increasing debt and fiscal manage- 100% 80% ment capacity (World Bank 2020e). In the standardized 60% project appraisal framework, road programs should 40% be evaluated and prioritized in a transparent, solid 20% manner, especially if projects are complex and large. 0% Additional attention must be paid when private sector admin costs Maint. Roads rehab bridges upgrade roads projects Tech. capacity Road safety PPP General & Construct & Other & birdges & studies Rehab & participation is involved. (See more in the following sections.) Internal sources External sources Avg. execution rate Note: Average execution rates are calculated by excluding potential outliers (greater than 150 percent). For public-private partnerships, annual execution rates are available only for 2015 to 2018. 16000 14000 of MT 12000 10000 70 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Increasing Efficiency in Road Procurement 111.  Unit costs for road projects can be decreased Figure 64. Unit Rates For Road Upgrading in Africa by increasing market competition in public procure- ment. While limiting expenditures strategically, it is important to minimize project costs and maximize effi- ciency gains in public procurement. There is a view that unit rates for road projects in Sub-Saharan Africa are increasing significantly (e.g., AfDB 2014). Although data are fragmented, unit rates seem to have increased, especially for larger road projects (figure 64). 112.  Road procurement costs normally benefit from economies of scale (AfDB 2014). In particular, unit costs for smaller contracts of 100 lane-km or less would be quite high; combining small lots into a fewer Source: AfDB (2014). bigger contracts could save public resources while minimizing project and contract management costs on the procurer side, although local contractors often Figure 65. Routine Road Maintenance Costs in Mozambique have a cost advantage in small contracts. Packaging 140 a large contract may disqualify local enterprises from 120 participating in competitive bidding, so it is important 100 to balance economies of scale with cost advantages of MT'000/km 80 local bidders. 60 40 113.  Unit costs of road projects seem to have 20 declined, at least for routine maintenance projects. 0 No systematic data are available, but this is considered 2015 2016 2017 2018 Routine Maintenance of unpaved roads Routine Maintenance of paved roads to be partly attributable to the economic slowdown Source: AfDB (2014). after the debt crisis. From 2015 to 2018, unit costs for routine maintenance declined 7 percent per year for paved roads and 17 percent for unpaved roads. Unit costs for unpaved roads declined from USD1,734 the road sector using output- and performance-based in 2015 to USD994 in 2018. The evolution of routine road contracts (OPRC) and community-based mainte- maintenance costs for paved roads was less regular, nance contracts. but costs declined from USD1,890 to USD1,501 per km over the period (figure 65).  In theory, OPRC can tighten an incentive 115. mechanism for contracting and reduce spending of  Because an economic recovery and rebound 114. public resources. It allows contractors to integrate dif- from the COVID crisis are expected in coming ferent activities, such as design, investment, and main- years, it will be of particular importance to con- tenance, into its final output, and deliver agreed-upon tain any increase in road project costs. To increase performance standards at minimum cost. Contracting economic efficiency in public procurement, the gov- authorities could also minimize administrative costs for ernment can explore and promote good practices in implementing and supervising projects. For instance, 71 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Lancelot (2010) showed that performance-based con- Routine maintenance is perhaps the best way to pre- tracts were 20 percent to 35 percent cheaper than serve existing road assets. Each dollar spent on routine traditional input-based road projects in Brazil. In other maintenance can save USD3 to USD5 on future reha- developed countries, reported savings range from 10 bilitation and reconstruction projects (ADB 2018). The percent to 30 percent (ADB 2018). Performance-based scope of maintenance activities contracted depends contracting can also ensure timely implementation on the complexity of projects and duration of contracts of necessary maintenance, ensuring road access to (figure 66). For routine maintenance on unpaved roads, communities during the rainy season and supporting 1- to 4-year community-based maintenance contracts livelihoods (Iimi 2020). can be used, which has the advantage of generating local jobs and meeting local maintenance needs in a 116.  In Mozambique, OPRC has been used in major timely manner. Each contract can cover 20 to 50 km. road rehabilitation projects, with a relatively short maintenance period. In Sub-Saharan Africa, the same 119.  For routine maintenance on paved roads, approach has been successfully implemented in Zam- microenterprise performance-based contracts bia and Liberia. Since the late 2000s, the government are useful, which can contribute to private sector of Mozambique has been exploring opportunities to development in local economies. Microenterprise use OPRC in its road programs. After several failed performance-based contracts can cover 50 to 200 km attempts, a hybrid form of OPRC was recently imple- for 4 to 10 years. For contracts longer than 10 years, mented, under which the government is responsible the entire scope of work and services can be covered for the design of rehabilitation projects, and the tradi- as a full-fledged OPRC. In Malaysia, an OPRC covers as tional input-based contract method is applied, whereas much as 7,000 km of roads (ADB 2018). Mozambique the private sector is fully responsible for the mainte- should explore these possibilities in road maintenance nance phase, under which work prices are capped. The projects, which will support the country’s objectives of perception of the private sector is that it is too risky to sustainable, inclusive growth and creation of well-paid take full responsibility for project design and availabil- jobs (World Bank forthcoming). ity of necessary materials in Mozambique. The mainte- nance periods are also relatively short (2.5 to 5 years). (See ANE 2014; 2020 for more detail.) Figure 66. Complexity and Duration of Performance-Based Road Maintenance Contracts 117.  As experience is accumulated, private sector responsibility under OPRC could be expanded. Con- tractors may be able to bear design and construction risks. To this end, close consultation with the private sector is important. The maintenance phase can be extended up to 30 years (ADB 2018). To ensure the long-term sustainability of OPRC, it is critical for the government to enhance high-level supervision capacity and performance monitoring, particularly during the maintenance phase.  Community-based maintenance contracts can 118. Source: ADB 2018. also be used more in small maintenance projects. 72 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Optimizing Expenditures Given Limited Resources 120.  Because available resources were nearly financial resources, this accounts for approximately halved, Mozambique must improve allocative half of total need. The evaluation was conducted based efficiency in the road sector. As discussed in the on road network inventory, condition, and traffic data previous sections, the debt crisis reduced road sector collected in 2015/16, estimating current network con- resources from more than USD400 million to approxi- ditions as of 2020 and assessing the financial need for mately USD200 million per year. The government used preserving it for the next 20 years.28 The estimated to devote more than half of its road sector resources 2020 conditions may not be accurate and are possi- to rehabilitating and upgrading primary roads. Road bly overestimated because the effects of the recent maintenance projects accounted for 20 percent to 30 cyclones that hit the southeastern African coun- percent. Such allocations are neither sustainable nor tries were not taken into account. Some roads were optimal to preserve the country’s current road assets. severely damaged. Thus, the results of this evaluation should be considered a rough estimate of the preser- 121.  The Mozambique paved road network was vation needs of the network. strategically evaluated using the Road Network Evaluation Tools (RONET), an Excel-based model 122.  Although capital unit costs vary according to used to assess the performance of road maintenance road network type because of their different standards and rehabilitation policies and determine proper allo- (table 8), routine maintenance costs are assumed to be cation of expenditures among various activities, such constant at USD2,000 per km per year based on past as recurrent maintenance, periodic maintenance, and expenditures by ANE. According to the RONET logic, rehabilitation road projects. See appendix B for more roads in good or fair condition (sustainable network) details on RONET. The evaluation covered only the require periodic maintenance, and roads in poor or part of Mozambique’s paved roads for which the nec- very poor condition require rehabilitation. Although essary data were available (6,677 km, approximately actual costs of road projects vary widely, the analysis one-quarter of the total network) (table 7). In terms of uses official estimated costs provided in ANE 2019. 28 RONET analysis requires more-detailed road condition and traffic data than visual road condition assessments, whose results are depicted in the previous sections. Table 7. Actual (2015/16) and Projected (2020) Road Conditions 2015/16 2020 International Roughness Index Network <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Total Primary or trunk 974 2,115 1,454 216 0 458 1,730 2,106 416 50 4,760 Secondary or main 104 394 446 128 0 54 130 572 316 0 1,072 Tertiary or local 11 261 280 293 0 0 176 283 386 0 845 Total 1,089 2,770 2,180 638 0 513 2,036 2,961 1,118 50 6,677 Sources: ANE; World Bank estimates. 73 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 8. Assumed Unit Costs of Road Projects (USD per km) Road Road project class Road project type Primary or Secondary Tertiary or condition trunk or main local Good Periodic Preventive treatment 29,250 27,000 22,500 Fair maintenance Resurfacing (5-cm overlay) 128,050 118,200 98,500 Poor Rehabilitation Strengthening (10-cm overlay) 256,100 236,400 197,000 Very poor Reconstruction 360,000 288,000 210,000 Source: ANE. 123.  Four scenarios are considered: Optimal sce- USD182 million would be needed per year for the first nario under which an optimal investment and mainte- 5 years, which is roughly equivalent to the amount nance solution is implemented without budget con- that the government spent annually from 2011 to 2018 straints (maximizes net present value of road assets), (figure 67). With administrative and other expenses budget constraint scenario with USD100 million excluded, total annual investment and maintenance (roughly equivalent to available resources after debt expenditures on roads and bridges was an average crisis), budget constraint scenario with USD60 million of approximately USD320 million per year for 2011 to (assumes further reductions of available resources), 2018, of which approximately 60 percent was spent on and minimum scenario with only routine maintenance national road investment and maintenance. With avail- (table 9). able resources, the government could have maintained the paved road network in the optimal manner. If the  During the first half of the 2010s, it is esti- 124. network is optimally maintained, approximately USD69 mated that the government had sufficient million would be needed annually in the following 15 resources to maintain the paved road network years. Under the optimal scenario, the maximum net optimally. Under the optimal scenario, to maintain the present value of the road assets could reach USD2,977 current paved road network (6,677 km), approximately million (figure 68). Table 9. Evaluated Budget Scenarios Scenario Description Optimal Implement optimal solution without budget constraints (maximizing net present value) Constraint (1): With currently available resources Spend USD100 million (average spending for last 3 years) Constraint (2): With more limited resources Spend USD60 million per year Minimum Only routine maintenance 100 Routine Annual costs ($ million p 80 60 74 40 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 20 0 Optimal Constraint Optimal Constraint Optimal Constraint 1 1 1 Primary Secondary Tertiary  To follow the optimal scenario, approximately 125. Figure 67. Estimated Annual Costs According to Scenario 70 percent of resources should be allocated to peri- 200 Rehabilitation odic and routine maintenance. This is very different Road agency costs ($ mil / year) 180 Periodic from the government’s actual allocation. As discussed 160 Routine 140 above, road maintenance accounted for on average only 120 30 percent of the total budget executed in the 2010s, 100 Actual which indicates significant under-maintenance of the 80 road network, as in other developing countries in Africa. 60 40 Budget 20 126.  With currently available resources, it is more 0 0 Yr 1-5 Yr100 200 6-20 Yr 1-5 300 Yr 6-20 Yr Yr 1-5 Yr500 400 1-5 Yr 6-20 6-20 challenging to preserve the existing road asset Rehab/paving national roads Rehab/paving regional roads value, possibly resulting in a 20 percent loss of the national roads Constraint 1 Maint.Optimal Constraint Maint. Do Minimum 2roads regional Maint. district/ municipal roads Emergency works current road asset value. As discussed, available road Bridge construction/ Rehab Maint. bridges resources have become even more limited. From 2016 Admin etc. to 2018, total investment in and maintenance expen- Figure 68. Financial Needs and Estimated Asset Values ditures on roads and bridges averaged approximately 3,500 USD160 million per year, which translates to a budget Optimal 3,000 Net Present Value (US$ mil) of USD100 million for the national roads. With this level Constraint 1 2,500 of budget (constraint 1), the maximum achievable road 2,000 -20% asset value would be USD2,379 million, 20 percent Constraint 2 lower than the optimal scenario. If available resources 1,500 are reduced further, for instance, by USD60 million, 1,000 there is a risk that half of the road asset value would 500 be lost for the next two decades. The do-minimum sce- Do minimum 0 nario would not preserve any road asset value. 0 50 100 150 200 Annual Road Agency Costs for Year 1-5 (US$ mil) 127.  Even under the constrained scenario, half of available resources should be spent on road main- tenance. This looks very close to the government’s Figure 69. Estimated Annual Cost According to Road allocation of road funds in recent years (2016–18). As Class and Scenario discussed in the previous section, maintenance expen- 140 Rehabilitation Annual costs ($ million per year) ditures accounted for nearly half of total executed road 120 Periodic spending in 2018. If this is the new normal, the current 100 Routine allocation between capital and recurrent expenditures 80 60 should be maintained, although it would still lead to a 40 poorer outcome than the optimal scenario. 20 0  Regardless of how much is available, the 128. Optimal Constraint Optimal Constraint Optimal Constraint primary road network is the first priority. Approx- 1 1 1 imately three-fourths should be spent on primary Primary Secondary Tertiary roads, which are more valuable and important to be preserved because they carry more traffic (figure 69).29 200 Rehabilitation cy costs ($ mil / year) 180 Periodic 160 29 The current RONET analysis focuses solely on the paved road network of Routine 6,677 km. 140 120 100 80 60 % of roads in good/fair condit 80 75 60 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 40 Optimal Constraint 1 20 Constraint 2 Do minimum 0 0 5 10 129.  With currently available resources for Figure 71. Road Length According Year to Road national roads (USD100 million for primary roads; Condition in Year 10 USD200 million for all roads), it is likely that road 8000 Very Good Good Fair Poor Very Poor Length of roads by road condition conditions will continue to deteriorate. Under the 7000 optimal scenario, the average international rough- 6000 ness index can be improved to 3.3, which can be sus- 5000 tained over time (figure 70), but it would increase to (km) 4000 5.1 under the constraint 1 scenario of USD100 million 3000 in spending. If only minimum routine maintenance 2000 is performed, the index would increase to 6.8 m/km. 1000 Approximately 80 percent of paved roads that have 0 Current Optimal Constraint 1 Constraint 2 Do been analyzed are in good or fair condition (figure 71). minimum Under the optimal scenario, that could be increased to nearly 95 percent, but under the constrained scenario (USD100 million), it would decline to 53 percent in 10 years, jeopardizing efficient movement of goods and people around the country. Investing in Climate Resilience Figure 70. Average Road Roughness Over Time 8 130.  Climate change increases the need for invest- 7 ment in the road sector. As discussed above, Mozam- Average roughness (m/km) bique is highly vulnerable to extreme climate events. 6 Climate change may complicate optimal road sector 5 spending. Although a comprehensive climate resilience 4 strategy should be implemented with road inven- 3 tory data updated from the resilience point of view, Optimal 2 Constraint 1 more investments are needed in resilient transport 1 Constraint 2 infrastructure (e.g., bridges and culverts). Resilient Do minimum infrastructure generally costs more. A global study 0 0 5 10 estimated that resilience would cost 3 percent to 23 Year percent more (World Bank 2017). In Mozambique, ANE has assessed the potential effect of extreme climate events and explored possible measures and found that 1 00 % of roads in good/fair condition resilient roads would add 10 percent to more than 100 80 percent to investment costs, depending on specifica- tion, placing additional fiscal pressure on the govern- 60 ment (table 10). 40 Optimal Constraint 1 20 Constraint 2 Do minimum 0 0 5 10 Year 76 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table 10. Comparison of Capital Costs for Rehabilitation or Reconstruction of Road Road type Road class Traditional standard Resilient standard Change (%) USD/km Gravel Primary 79,763 87,600 9.8 Secondary 49,800 81,600 63.9 Tertiary 25,149 60,000 138.6 Paved Primary 298,800 500,000 67.3 Secondary 239,040 400,000 67.3 Tertiary 174,300 350,000 100.8 Source: Bank estimates. 131.  Resilient roads are more valuable but require greater capital investment. For simplicity, if building resilient roads increases the unit costs of road projects uniformly by 20 percent (table 11), RONET estimates that the required investment cost would increase to USD196 million per year under the optimal scenario (figure 72). More rehabilitation projects would be required than in the baseline case. The maximum possible road asset value would also increase to USD6,618 million (figure 73), although average road conditions would be slightly worse than in the baseline case (figure 74). Under the climate change scenario, it would be more difficult to maintain the road network, which would make it more vulnerable. Preserving the road network is more important than ever (figure 75). Table 11. Comparison of Capital Costs for Rehabilitation or Reconstruction of Road Road condition Road project type Road work type Primary or Secondary Tertiary or Trunk or Main Local Good Periodic maintenance Preventive treatment 35,100 32,400 27,000 Fair Resurfacing (5-cm overlay) 153,660 141,840 118,200 Poor Rehabilitation Strengthening (10-cm overlay) 307,320 283,680 236,400 Very poor Reconstruction 432,000 345,600 252,000 Source: Bank estimates. 80% Routine Annual costs by work type 70% maintenance 60% 77 50% Periodic maintenance INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 40% 30% Rehabilitation 20% 10% 0% um al t1 t2 tim in in im tra tra Op in ns ns m Co Co Do Figure 72. Annual Costs: Baseline and Resilience Figure 73. Road Asset Values: Baseline and Resilience 250 7000 Rehabilitation mil / year) 7000 Periodic 6000 200 Value (US$ mil) Routine 6000 5000 150 mil) (US$ ($ 5000 4000 100 condition Valuecosts 100 4000 3000 Net Present 80 Road agency 50 2000 % of roads in good/fair 3000 Net Present 0 60 1000 2000 Yr 1-5 Yr 6-20 Yr 1-5 Yr 6-20 Yr 1-5 Yr 6-20 Yr 1-5 Yr 6-20 400 Baseline - Baseline - $100 Resilient - Resilient - $100 1000 0 50 100 150 200 250 Optimal mil Optimal mil Baseline - Optimal Baseline - $100 mil 20 Annual Road Agency Costs for Year 1-5 (US$ mil) 0 Baseline - Do minimum Resilient - Optimal Resilient - $100 mil 50 100 0 150 200 250 0 Annual Road Agency Costs for Year 1-5 (US$ mil) 8 0 5 10 Year Figure 74. Average Roughness Figure 7 75. Length of Roads According to Condition (m/km) 8 6 8000 Very Good Good Fair Poor Very Poor condition 5 7000 roughness 7 Average roughness (m/km) 4 6000 6 3 5000 by road 5 Average 2 4000 (km) Baseline - Optimal Baseline - $100 mil 4 Length of roads 1 3000 Baseline - Do minimum Resilient - Optimal 3 2000 Resilient - $100 mil 0 1000 0 5 10 2 Baseline - Optimal Baseline - $100 mil 0 Year 1 Baseline - Do minimum Resilient - Optimal Resilient - $100 mil Current Baseline Resilient Baseline Resilient Do 0 minimum 0 5 10 Optimal $100 mil Year 132.  Pavement is among the most important 133.  Resiliency standards may have to be consid- measures of resilience. A systematic paving deci- ered in more road projects. A recent road project sion model, which is a multicriteria analysis based on was planned in Mozambique to rehabilitate eight socioeconomic and engineering factors in five cate- road sections. According to normal road standards, gories (country context (e.g., income level and disas- four sections were identified to be paved with a full ter exposure), subnational context (e.g., poverty and specification of double bituminous surface treatment agricultural production), operational environment (e.g., (DBST). When a systematic paving decision model was road asset management system and design stan- used, two more segments were identified as needing dards), road characteristics (e.g., traffic and access to to be paved or sealed because of the potential need social facilities), and engineering factors (e.g., drainage for essential connectivity and productive use of these and availability of materials)), could be used to deter- roads. More resilience is needed to build better. mine whether a road should be paved. Based on the weighted average score, potential need for pavement is assessed. 20 Fuel levy on petrol ($ Recommended = 17 cents 78 15 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: cents/liter) TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 10 Avg. = 8.9 cents 5 0 Increasing and Diversifying Road Sector Revenues B e Bu en in biq l i T a Cô e da K so nz i Gh ep . oz M e e ad Ch . M i bia Le i a an al d’l on S ie Eth ogo o, r ag ad Ga ea-B inea in u r Za n di Co N b ia Na eny a Ca w an ia Af S en ire i a, u p Ta al aw ng i ge Gu G sca am a Th u an on mb i ssa ric eg R an Re rr a i op Fa te r o vo R m ru m a na m rk i M Bu M ral nt Ce 134.  With limited resources in the road sector, it Figure 76. Road Fund Resources in African Countries (USD/km) is difficult to achieve optimal road asset manage- 10000 ment. As discussed above, it is essential to optimize RF resources ($/km) 8000 use of available resources, but even if all resources are 6000 used in the best way, it may not be possible to maintain 4000 the existing road network, as shown above; available resources are simply insufficient. Road fund revenue 2000 per km of road is approximately USD9,200 in Mozam- 0 bique, which is the highest in the region, because S ie B al i a da M a so M i Co Gu n e na ia M i a, i a Ta Togo Bu Nam on o, a Rw scar Ga Ethi n di Ca Ken a Za ia me y a Cô S en u e Le n Gh ire Ch . ag e d’l al Bu er Ni d w p i ng ine an rr a en i a ad Th rk i i b an mb op te eg mb Re g al a o an vo o iq ru F r a nz mb the Road Fund consolidates external and internal a oz M resources (figure 76). Still, it covers only routine main- Source: World Bank 2019b. tenance and possibly some periodic maintenance. 135.  For long-term sustainability, it is ideal to Figure 77. External and Internal Resources for Road Fund reduce dependence on external resources. Mozam- 25,000 70 62.1 bique relies, on average, approximately 60 percent 57.8 62.0 Resources collected (MT million) Share of internal resources (%) 50.5 50.9 60 20,000 48.6 on external resources for road sector expenditures, 45.2 50 which has been increasing in recent years (figure 77). 15,000 35.0 40 External resources are volatile and tend to be focused 10,000 30 on investment projects. In addition, the execution rate 20 5,000 of the externally financed budget has been declining 10 (figure 78). Over the long run, internal revenues that 0 0 2010 2011 2012 2013 2014 2015 2016 2017 2018 can be dedicated to the road sector must be increased and diversified. External resources Internal resources Share (%) 136.  Mozambique must increase internal 35,000 Road fund budget Internal resources 5.0% resources, which are generally more stable and Figure 78. Execution of External (%of GDP) Resources (% of GDP) 4.5% (MT million) 30,000 sustainable. The Road Fund’s main internal resources 4.0% 25,000 25,000 160% 3.5% are Treasury subsidies and road user charges. There (MT million) 140% 20,000 20,000 3.0% are some bank loans and own earnings, but these are 120% budget Execution rate (%) 2.5% 15,000 100% negligible. Government regulations allow the Road 15,000 2.0% resources 80% Fund to use 17 potential resources,30 of which five road 1.5% fund 10,000 10,000 60% user charges are currently used: levies on diesel, gaso- 1.0% External 40% Road 5,000 5,000 0.5% 20% line, or other energy sources used for motor vehicles; 0 0 0.0% 0% international transit fees; tolls; border crossing fees; 2010 2011 2012 2013 2014 2015 2010 2011 2012 2013 2014 2015 2016 2017 2018 2016 2017 2018 External Financing (planned) External Financing (actual) Execution rate (%) and road concession taxes. The fuel levy accounts for 90 percent of total road user charges in Mozam- bique (figure 79). In theory, fuel tax revenues increase as people drive more. Thus, they are sustainable resources, allowing road authorities to maintain the road network sustainably. 30 Decreto no 61/2019. Ajusta as atribuçaões, gestão, regime orçamental, tutelar, organizacão e funcionamiento de Fundo de Estradas, criado pelo decreto no 22/2003. 79 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Figure 79. Composition of Internal Resources Figure 80. Fuel Levies in African Countries 1.5% 0.1% 20 Fuel levy on petrol ($ Recommended = 17 cents 15 cents/liter) Road user 10 Avg. = 8.9 cents charges 52.7% 59.5% 5 39.0% 0 B e Bu en in biq l i T a Cô me da K so nz i Gh ep . oz M e e ad Ch . M bia Le i a an al d’l on S ie Eth ogo o, r ag ad Ga ea-B inea in u r Za n di Co N b ia Na eny a Ca w an ia Af S en ire i a, u p Ta al aw ng i ge Gu G ca am a Th u an on mb i ssa ric eg R an Re rr a op Fa te r o vo mi s R m ru a i 0.8% na 6.0% rk i M Bu M ral nt Ce Treasury subsidy Bank Loans Own earnings Tax on fuel Highway tax Toll charges Source: World Bank 2019b. 10000 RF resources ($/km) 12,000 8000 140% 6000 120% 10,000 137.  Fuel taxes are low and have potential to be 138.  The fuel levy is only partly ring-fenced, vio- 4000 100% Millions of MT increased. In the short run, Mozambique should 8,000 lating the principle of road funds. In addition, the 2000 80% increase 6 ,000 fuel levies, which consist of fixed tariffs on collected fuel levy is not perfectly transferred to the 0 6 0% each liter of diesel and gasoline distributed in the Road Fund. Only 75 percent and 50 percent of fuel levy S ie B al i a da M a so M i Co Gu n e na ia M i a, i a Ta Togo Bu Nam on o, a Rw scar Ga Ethi n di Ca Ken a Za ia me y a Cô S en u e Le n 4,000 Gh ire ag e Ch . d’l al Bu er Ni d w p i ng ine an rr a en i a ad Th rk i i b an mb op te eg mb country. Fuel taxes are approximately 3 meticals, on diesel and gasoline are ring-fenced, respectively. Re g al a o an 40% vo o iq ru F r a nz a mb or approximately 5 U.S. cents per liter (table 12), 2,000 The Road Fund 20% relies on a Treasury subsidy for annual oz M which 0 has not been adjusted since 2003, meaning resources, which 0% are subject to political arbitration. that implicit 2011 for 2010subsidies gasoline 2012 2013 and diesel2015 2014 have 2016 This 2017internal 2018 resource structure may undermine the increased substantially Total Planned over the last Internal Funding two Total decades Achieved Internal Funding RoadRelease Fund’s independence and accountability for pre- rate (Governo de Moçambique 2015). The regulation allows serving the road network objectively. It is not aligned trimestral adjustment to accommodate inflation, but it with the main principle of second-generation road does not seem to have been implemented for a long funds (Benmaamar 2006). If 100 percent of fuel levies time. As a result, the current fuel levy in Mozambique were allocated to the Road Fund, financial sustainabil- is below the Sub-Saharan African average, which is 8.9 ity and planning reliability could be improved. U.S. cents, and further below the recommended level of 17 U.S. cents per liter to fully address road mainte- 139.  Revenue sources other than the fuel levy nance needs (figure 80) (World Bank 2019b). Fuel tax should also be explored further. Given current and rates should be increased and kept adjusted to accom- expected technological advances, fuel levies are not modate general changes in domestic prices. a sustainable revenue source for the road sector. As motor vehicles become more efficient and the fleet is renewed toward hybrid or even electric vehicles, Table 12. Fuel Tax Rate and Allocation in Mozambique revenues from fuel taxes will plateau or even diminish, as observed in advanced economies. Different modal- Type of fuel Diesel Gasoline ities to charge road users must be considered. Con- Tariff per liter (meticals) 3.02 3.29 ceptually, there are eight categories of road funding instruments (table 13). Some are more implementable Ring-fenced to Road Fund (%) 75 50 than others for technical and socioeconomic reasons, such as affordability, user acceptance, yield potential, enforceability, and administrative simplicity (World Bank 2019b). 80 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 140.  Distance-based charges, such as heavy goods most damage to roads. In the medium to long term, vehicle charges, may be of particular interest Mozambique could investigate the potential to charge in Mozambique. Although fuel levies are the most heavy goods vehicles based on distance or a mix of practical instrument in the short run, distance-based distance and load on selected highly trafficked align- charges, such as heavy goods vehicle charges, should ments. There are a number of successful examples, be investigated as a next step. Despite general public such as Poland after they joined the European Union. opposition, often experienced when increasing fuel lev- There are many prerequisites to introduce such a sys- ies, this instrument has many desirable features, hav- tem, for instance, communication campaigns with the ing relatively good yield potential and being simple to trucking industry and a sustained mechanism to ensure implement. It also reflects the user-pays principle to a the high quality of the alignments to be charged. large extent, given that heavy goods vehicles cause the Table 13. Categories of Road Funding Instruments Type of funding instrument Description General all-purpose taxes No direct link to road network use or development. May consist of direct (e.g., income tax) or indirect (e.g., value-added tax) taxes. Nonrecurrent access-based One-time charges (e.g., registration fee, customs duty for imported vehicle, luxury car tax) at time of motor charges vehicle purchase to allow use of motor vehicle on road network. Recurrent access-based Vehicle licensing fee, axle tax, insurance contract tax, driver’s license fee. charges Fuel-consumption based Excise taxes included in price consumers pay at the pump; taxpayer is oil companies or distributor. Consist charges of a unitary tariff (e.g., USD/gal) that can be modified depending on fuel type. For example, less-polluting fuels such as liquefied petroleum gas and compressed natural gas may have a lower rate. The fuel levy, which accounts for an average of 80 percent of the Road Fund’s resources, is usually carved out of the fuel excise tax. Distance-based charges User pays directly in exchange for the use of road. Usually based on unitary tariff (e.g., USD/mile traveled). Can be modified according to several parameters (e.g., maximum authorized weight, number of axles, time of day, greenhouse gas emission class). User can pay with cash or electronic fund transfer. Charges include tolls (on specific stretches and linked to repayment of road investment, e.g., in user-pays public-private partnership scheme), international transit fees, heavy goods vehicle charges (on a network basis; not necessarily ear- marked to repay a specific investment). Time-based charges User pays directly in exchange for right to use road (or network) during a specified period. Usually based on unitary tariff (e.g. USD/day, week, month, or year). Can be modified according to several parameters (e.g., maximum authorized weight, number of axles, time of day, greenhouse gas emission class). Charges include vignettes and urban charges. Universal road user charge Recognizing the limits of the current road funding system, road maintenance and investment gaps, and future shortcomings of fuel taxes as one of the main funding instruments, some countries have started considering a new road pricing scheme that would first complement and then replace current funding instruments. Related charging arrangements would apply to all motor vehicles and the entire road network. The concept is basically to charge users for the distance they travel on roads. In addition to distance, pricing could include vehicle weight class, time of day, location, type of fuel, or greenhouse gas class. Value capture charges Public recovery of all or a portion of property value created because of public infrastructure investment or ben- efits from it. Subject to enabling conditions (e.g., real estate market vitality, zoning and land use entitlements), new road capacity and new road access can create business opportunities and value in surrounding land and real estate. They include impact fees, special assessment districts, and tax increment financing. Source: World Bank 2019b. 81 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING  In theory, PPPs can ease fiscal pressure on the 141. 106 km is in Mozambique (table 14). There are two toll government, but their potential must be examined plazas—at Moamba and Maputo. The current tolls are carefully. Mozambique has one of the few successful low for the Mozambique part—approximately USD3 PPP road projects in Africa—National Highway (N4), for light vehicles and USD20 for heavy vehicles with connecting Pretoria to Maputo. In 1997, a 30-year con- three to four axles when a vehicle drives the entire cession was agreed upon with Trans African Conces- stretch of 106 km (table 15). There are other road and sions, which involves an investment of approximately bridge concessions in Tete and Maputo and along the USD420 million to manage a 570-km road, of which Beira corridor. Table 14. Road Funding Instruments Name Road Year Length (km) Investment Duration Annual average daily (USD million) (years) number traffic Trans African Concessions: Maputo N4 1997 570 (106 in 426 30 1,834 Corridor to South Africa Mozambique Estradas do Zambeze: Tete bridge and 2,555 9,447 34 746 780 7.0 regional highways connecting Malawi, Zimbabwe, and Zambia to Beira Maputo Ring Road and Ka Tembe 2014 311 785 4,000a Bridge Beira to Zimbabwe border at 2,515 11,058 35 874 909 8.1 Machipanda Sources: PPIAF 2009; ANE 2020. a. Projected. Table 15. Trans African Concessions Toll Fees Vehicle class Toll at plaza (meticals) Total cost per km Total for Moamba Maputo 106 km Meticals U.S. dollars Light 190 35 225 2.12 0.03 2-axle heavy 480 120 600 5.66 0.08 3-4 axle heavy 1,100 350 1,450 13.68 0.19 >5 axle heavy 1,500 500 2,000 18.87 0.27 Source: ANE 2020. 82 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 142.  PPP viability in roads and highways depends Figure 81. Distribution of Road Networks According on current and future traffic. As is well documented to Traffic Level in the literature, there is a substantial risk of traffic in 100% toll road PPPs. In the United States, toll revenues were 80% 20 percent to 75 percent lower than original forecasts. 60% > 10000 Worldwide, a wide range of errors in traffic forecasts 40% 3000-10000 can be found: The ratio of actual to forecast traffic 20% 1000-3000 ranges from 0.14 (i.e., 86 percent lower than expected) 300-1000 0% 100-300 to 1.5 percent (50 percent more than expected), with Zambia (only national Mozambique Zimbabwe South Africa Tanzania Sub-Saharan Africa avg. (all roads) < 100 a mean of 0.77 (i.e., 23 percent lower than expected) roads) (PPIAF 2017). Sources: ANE (2020) for Mozambique; Gwilliam (2011) for other countries. 143.  Given current traffic, the opportunity for toll- 100% based PPPs is limited in Mozambique. Global expe- rience shows that the minimum threshold for user 80% Figure 82. Annual Average Daily Traffic payments or toll-based PPPs for road projects may be (Number 60% of Vehicles/Day), 2017 > 10000 an annual average daily traffic (AADT) of approximately 40% 3000-10000 1000-3000 20,000. In Mozambique, the distribution of road net- 20% 300-1000 works according to traffic level is highly skewed. Most 0% 100-300 No of links AICD Link only National KM of links of the national roads have less traffic than the thresh- < 100 old (figure 81). There may be some potential for toll- based PPPs along N7 from Tete to Zobue (Malawi bor- der), N1 from Nampula to Nacala, and N1 from Maputo to Xai-Xai (figure 82), but traffic levels are generally low (5,000–15,000 vehicles per day). There are only a few opportunities for toll-based PPPs, at least in the short to medium term. 144.  Different types of PPP arrangements are available, but good governance is critical. There are different types of PPPs in roads and highways, depending on private sector responsibility and fund- ing sources (figure 83). Even if not enough traffic is expected to attract private sector investment, avail- ability-based payments could fill a viability gap. In addition, risk sharing can be rebalanced to make it bankable, although good governance and transpar- ent public investment management are essential for a successful PPP. The debt crisis in 2016 highlighted this. As global experience indicates, road-sector PPPs are Source: Based on ANE data. highly sensitive to future traffic forecasts. Although it is too early to evaluate the performance of the current 83 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING PPP in Mozambique, it has been seen that commercial 146.  To expand PPPs and OPRCs, Mozambique must viability is highly challenging for some PPPs in which increase its institutional capacity. It has recently the initial investment exceeded hundreds of millions of adopted PPP regulations. ANE (2020) has prepared a dollars but current traffic is very modest. A solid, trans- preliminary feasibility study for PPP programs in the parent public investment appraisal system must be road sector, although it is unclear whether and how established, possibly alongside the ongoing macro-fis- the government appraises and supervises PPP trans- cal management initiative, Gestao Economica para actions. Details are unclear for one of the successful Desenolvimento Inclusivo (World Bank 2020e). concessions in Mozambique (N4 Maputo corridor), let alone for other PPP concessions. The appropriate Figure 83. Typology of Toll-Highway governance structure must be put in place to address Public-Private Partnership Projects unsolicited proposals and advance preparation pro- Figure 764. Typology of Toll-Highway Public-Private Partnership Projects Revenue source cedures in a competitive, transparent manner (World User Operating Build and Bank 2018). A diagnosis of road PPP readiness is a use- concession Operate Full Concession payments, tolls (Lease) Concession (BOT) ful policy tool to identify improvement areas and how to unleash the potential of road PPPs. Design, Build, Design, Build, Government Management Operate, Finance, payments, Contract Maintain Operate & subsidy (DBOM) Maintain 147.  The road restoration window model—a mech- Role of the anism that dedicates resources to road restoration private sector Operations Construction Construction, only & operations operations, finance contracts that can be established within the Road Fund Source: PPIAF 2017. —may be a way to balance affordability for users with financial sustainability. These contracts, which the public or private sector can finance, would be long- 145.  The main objective of PPPs is to increase effi- term performance-based contracts bundling an initial ciency in road programs, not to generate tolls or phase of road restoration, including resiliency projects, other revenues. In principle, there is no way for the with a subsequent phase of operation and mainte- government to escape financial obligations to invest in nance. Resources, which are gathered through tolls or and maintain the road network as long as it is a pub- other forms of distance-based charges on some of the lic good. Even though the private sector is involved selected restored roads, are used to fund availability in any form of PPP, road users and taxpayers bear and performance payments during the maintenance the costs of road development—now or in the future. phase. Any surplus would cross-subsidize the Road Comparing traditional road projects and PPP projects Fund’s existing road programs (World Bank 2019b). in Europe, the risk premium of cost overruns in a road Step-by-step capacity building is needed to establish concession is estimated at 24 percent of total project a road restoration window mechanism. There are also cost (Blanc-Brude et al. 2006). In U.S. highway con- prerequisites to establish such a mechanism, such as tracts, the implicit adaptation cost reflected in bids at in-depth knowledge of the road network (including 7.5 percent to 14 percent of winning bids or USD2.1 to traffic and road conditions), raising fuel taxes, legal USD2.4 per USD1.0 of expected contract adjustment and institutional set-up for PPPs, functioning road (Bajari et al. 2014). Thus, it is important to develop and charging policy, administrative capacity in ANE and the apply a systematic assessment framework to maximize Road Fund, and the private sector’s capacity to under- efficiency gains in public investment management, take OPRCs (box 5). including new contracting methods, such as OPRCs, to tighten incentive mechanisms for contactors. 84 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Box 5. Checklist of Activities to Perform to Implement the Restoration Concept Surveys and data analyses Upstream implementation phases - Independent survey and assessment of national road network to improve knowledge of road conditions and traffic levels and select priority alignments to be restored and, eventually, potentially tolled - Appraisal of road contracting industry to assess local contractors’ ability to participate in performance-based contracts, output- and perfor- mance-based road contracts, and road restoration public-private partnerships (PPPs) - Appraisal of financial industry to identify local commercial banks and institutional investors (e.g., pension funds) and assess their capacity to provide long-term financing and any existing national infrastructure bank and assess its capacity to address long-term financing gaps - Legal and institutional gap analysis to identify areas for improvement in the road asset management ecosystem; the Road Fund legal mandate, structure, and legal status (e.g., moving from an administration to a state-owned enterprise); the PPP framework; and the public procurement frameworka Institutional and legal reforms as appropriate Upstream phases after the outcomes of data analyses - Road Fund reform (e.g., changing Road Fund legal status, creating a road restoration window, enabling funding of spot reconstruction bun- dled with maintenance in long-term contracts - Financial sector reform (e.g., enabling local institutional investors to provide long-term financing for public infrastructure projects, reforming or restructuring an existing national infrastructure bank) - Toll policy elaboration - PPP framework adjustment as required Capacity building Upstream and midstream phases of implementation. - Road contractors Advertise and explain the Restoration Concept, organize classroom and on-the-job training programs on perfor- mance-based contracts, output- and performance-based road contracts, and road restoration PPPs to increase their capacity to qualify and bid successfully - Financiers Advertise and explain the restoration concept, improve their project finance skills and capacity to provide long-term financing - Public sector institutions Assistance to road funds (e.g., management of resources, disbursement processes, reporting and auditing) and road agencies (RAs) (e.g., additional skills and staff needed to manage PPP contracts) Road restoration PPP programming and implementation Midstream phases of implementation - Fundraising for advisory services Prepare and manage application for Global Infrastructure Facility (see https://www.globalinfrafacility.org/) funding, which may support programming and transaction implementation activities and is compatible with advisory mandates executed by an independent advisor or the International Finance Corporation - Preparation of a pipeline of priority alignments to be restored using road restoration PPPs based on surveys and data analyses - Drafting of standardized tender documents (request for qualifications, request for proposals, project agreement, direct agreement, funding agreement, toll agreement, interface agreement); - Market sounding to collect private sector stakeholder feedback on standardized tender documents, improve them as appropriate, and iden- tify market gaps requiring World Bank Group financial instruments (e.g., payment guarantees, political risk insurance) - Tendering road restoration PPPs If market sounding reveals that World Bank Group financial instruments will be necessary to support bank- ability, these instruments should be made available as early as possible in the tender process. Source: World Bank 2019b. a PPP, public-private partnership. 85 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Section Summary  Key takeaways from this section are as follows. 148. • The current financing mechanism in Mozambique’s more in public road procurement opportunities, road sector cannot maintain the road network, public procurement costs in the road sector could especially given a sharp reduction in internal be reduced as a whole. resources in recent years. • Within the road sector envelope, the government • To maintain the current road network, approxi- should allocate more resources to road mainte- mately USD400 million would be needed, of which nance while maintaining rehabilitation and upgrad- USD180 million must be allocated to the paved ing expenditures on the primary road network. network. This is equivalent to the amount that the • Building climate resilience into road infrastructure government devoted to the road sector before the is a wise investment in Mozambique. More upfront debt crisis in 2016, but available resources have investment could result in higher net present value been halved to approximately USD200 million. of road assets, although average road conditions • Given limited resources, the government faces may still be subdued (compared with the no-cli- renewed challenges to maintain the existing road mate-change scenario). network and ensure mobility to enhance inclusive • Road-sector revenues must be increased. In partic- growth. ular, fuel levies, which have been fixed for a long • Unnecessary expenditures must be avoided, and time, must be increased and kept adjusted, possibly strategic prioritization is necessary. The govern- up to a regional norm (10 U.S. cents per liter). ment has developed a prioritization tool, which • The fuel levy should be dedicated more to road sec- should be improved and applied strategically to tor development. The fuel levy is only partly ring- make road investments more selective. fenced, undermining the principles of road funds • Available resources must be used wisely; the gov- such as independence and sustainability. ernment must increase efficiency in budget execu- • Toll-based PPP opportunities may be limited, tion. Currently, the execution rate is approximately but other types of PPP arrangements could be 40 percent to 60 percent. It is important to build explored. The readiness of Mozambique to imple- a more solid, systematic project assessment and ment a road restoration window could be investi- monitoring system, possibly alongside the ongoing gated. Such a program would start with enhanced high-level macro-fiscal management improvement revenues from fuel taxes as one prerequisite. Then initiative (Gestao Economica para Desenolvimento selected priority alignments would be restored Inclusivo). A medium-term budgetary framework, using long-term performance-based road contracts, with more realistic project planning, especially for potentially leveraging private sector participation. donor-funded projects, must also be developed. Then some of the restored priority alignments • Efficiency gains in public procurement must be could be charged. Revenues from this charging made by introducing new contractual methods, system would be allocated to the Road Fund and such as OPRCs and community-based maintenance would fund availability and performance payments contracts. These mechanisms are expected not only of such road restoration contracts. Surplus reve- to help contain public procurement costs, but also nues would cross-subsidize existing road programs, to generate local jobs and facilitate local business enhancing revenues available to the Road Fund development. By strengthening market competition while gradually improving and sustaining road con- and encouraging local contractors to participate ditions in Mozambique. 86 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 6. Annexes 87 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Annex A. Literature Review Figure A.1. Results Chain Access to Access to Income and Physical Access to Local market education and input/ output welfare connectivity labor market development health markets increase 1-2 years 3-5 years 5-10 years Source: World Bank.  Greater transport access can bring a wide range 149. wider samples in Vietnam. In rural Tocantins, Brazil, of economic and social benefits in the short and long increasing road accessibility increased girls’ school term. Different studies discuss different types of attendance by 3 percent, but no significant effect was effects using different methods. The observed effects found for boys (Iimi and Lancelot forthcoming). tend to be highly heterogeneous and dynamic. The emerging results chain in the literature is shown in 152.  There is little evidence of any effect on health figure A.1. care access, let alone health conditions. Invest- ment in roads no doubt increases physical accessibil- 150.  Rural road investment increases physical con- ity to health care facilities (e.g., Danida 2010), but it is nectivity of rural residents. In rural Nicaragua, road unclear how greater rural access contributes to health. improvements in tertiary roads reduced travel time No effect on health conditions of household members to health care facilities by 5.8 minutes per km (Danida was found in Tocantins, possibly because health condi- 2010). By rehabilitating bridges and culverts, distance tions were not bad before the project (Iimi and Lance- and travel time to the district center were reduced by lot forthcoming). No difference was found in mortality more than 30 percent in Tocantins, Brazil (Iimi and in villages in Orissa, India, with and without all-weather Lancelot forthcoming). Improving the physical condi- roads, although treatment households reported tion of roads reduced the time required for an ambu- switching from care in primary health clinics to hos- lance to arrive in Georgia (Lokshin and Yemtsov 2005). pitals further away (Bell and Van Dillen 2012). One positive indication was found in Liberia; more people 151.  Improving rural roads is likely to increase visited health care facilities where local road connec- access to education, especially for girls. Many tivity, measured according to density, was better (Iimi studies have explored the effect of improved roads on and Kulwinder 2018). school attendance and school completion rates. Evi- dence is generally supportive, with some heterogene- 153.  Whether lower transport costs result in lower ity. In Bangladesh, rural road improvement increased transport prices depends on market structure. In secondary school enrollment for boys and girls, with Bangladesh, rural road improvement decreased aver- estimates ranging from an increase of 8 percent to age household spending on transport by 40 percent 38 percent depending on study area (Khandker et al. during the rainy season (Khandker et al. 2009; Khand- 2009). In rural Vietnam, primary school completion ker and Koolwal 2011). In Georgia, average transport improved by 30 percent, and secondary school enroll- spending declined (albeit not statistically significantly ment increased by 7 percent (Mu and van de Walle so) by 4.4 percent (Lokshin and Yemtsov 2005). In the- 2011). By contrast, Cuong (2011) did not find any sta- ory, consumers can benefit from lower road user costs tistically significant effect on school attendance with as long as the transport market is competitive. 88 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 154.  Improved road conditions can make new  With greater access to input and output mar- 157. transport services available, and people’s trans- kets, farmers are likely to choose to grow more port modal choice will change over time. Improved marketable crops. Among other determinants of crop road conditions seem to increase individual vehicle choice such as availability of irrigation and improved and motorcycle ownership. After the project, more seeds, transport access is important. In Uganda, house- people began to use individual vehicles and public holds’ selection of commercial varieties of bananas buses in rural Brazil (Iimi and Lancelot forthcoming). was found to decrease with distance from the market In Nicaragua, motorized transport intensity (especially (Edmeades et al. 2008). In Tanzania, farmers chose passenger vehicles and motorbikes) increased sig- food crops, such as maize and rice, for which transport nificantly after a road improvement project (Danida costs to the domestic market were low. Farmers are 2010). Although in a more urban setting, asphalting of more likely to choose export crops, such as cotton and residential nonarterial streets increased motor vehicle tobacco, where international market connectivity is ownership by 40 percent in Mexico (Gonzalez-Navarro better (Iimi and Humphrey forthcoming). and Quintana-Domeque 2016).31 158.  Land prices will increase, reflecting greater 155.  Motorized transport modes can generate land productivity. Rural road investments can make dynamic socioeconomic benefits from rural road areas that were too remote more profitable to culti- investment. Motorized and nonmotorized transport vate. In Nepal, plot values were found to increase with can generate much different effects. In Peru, a road proximity to the market, measured according to travel rehabilitation project caused an average increase in time (Jacoby 2000), although in rural Nicaragua, no household income of 35 percent in households on effect was found on land value (Danida 2010). Not only motorized roads, but no effect was found on nonmo- physical infrastructure, but also institutional factors torized roads (Escobal and Ponce 2002). In rural Bra- matter, such as structure of land markets and land reg- zil, improved road conditions allowed public school istration systems. buses to operate in remote areas, resulting in greater distance and travel time to school, but significantly 159.  Over the long term, greater transport con- greater quality of education services (Iimi and Lancelot nectivity can create new jobs in the agricultural forthcoming). and nonagricultural sectors. There is a debate as to whether rural roads create jobs in the agricultural or 156.  Agricultural input and out prices are a main nonagricultural sector. In Nicaragua, a rural road proj- way that farmers can benefit from rural road ect increased self-employment in the agricultural sec- investment. Greater road connectivity reduces agri- tor (Rand 2011). This is interpreted to mean that labor cultural input prices and increases output prices. In demand for agricultural workers increased as a result Bangladesh, fertilizer prices declined by 3 percent to of more favorable input and output prices enabled by 10 percent, and agricultural output prices increased transport cost reductions. Greater rural connectivity by 2 percent to 22 percent (Khandker et al. 2009). In can also increase access to other job opportunities. In Orissa, India, output prices increased 5 percent more Bangladesh, agricultural and nonagricultural employ- in villages receiving all-weather roads than in compari- ment increased (Khandker et al. 2009). In Vietnam, son villages (Bell and Van Dillen 2012). Cuong (2011) found an increase in total working hours. 31 This study is one of the few in the transport sector in which an intervention, street asphalting in this case, was randomly assigned. Thus, the result should be robust by construction. 89 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING 160.  Lower transport costs can bring new busi- average household consumption increased 16.3 per- nesses to communities. Greater rural accessibility cent (Dercon et al. 2008). As discussed above, in Peru, can create new jobs in communities. In rural Nicara- a road rehabilitation project increased household gua, the vast majority of new jobs were created in the income by 35 percent along motorized roads (Esco- agricultural sector, but the share of nonagricultural bal and Ponce 2002). In Vietnam, road rehabilitation employment also increased, for instance in the con- increased household income by 9 percent on average tracting sector (Danida 2010). In Vietnam, services that (Coung 2011). In Tocantins, Brazil, improved roads had not been available, such as hair dressing, became helped increase household monthly income by approx- available after the road rehabilitation project (Mu and imately 20 percent (100 reals, approximately USD67) van de Walle 2011). Local businesses were stimulated, (Iimi and Lancelot forthcoming). although it may take a long time to develop local busi- ness. It depends on urbanization and agglomeration,  Few studies have explicitly analyzed the effect of 162. which often follow nonlinear development paths (e.g., roads on the incidence of poverty. Dercon et al. (2008) Cieślik and Ryan 2004; Milner et al. 2006). is among the notable exceptions. In Ethiopia, poverty declined by 6.9 percent. Other studies, such as Khand-  All these changes in turn increase household 161. ker et al. (2009), provide an implicit indication that road income, reduce poverty, and improve people’s investments are pro-poor, meaning that economic welfare. The literature is robust about this. In Bangla- benefits are disproportionately higher for the poor desh, average household expenditures were 9 per- than for the non-poor. In Bangladesh, the most signif- cent to 10 percent greater in treated than comparison icant effect of a rural road investment was found for households (Khandker and Koolwal 2011). In Ethiopia, the 15th percentile of the sample. Table A.1 Study Context Intervention Economic effects Education effects Health effects Jacoby (2000) Nepal (rural) Natural variation 27%↑ in land value in travel time to 4.7%↑ in agricultural wages market Escobal and Peru (rural) Rural road 35%↑ in income for house- Ponce (2002) rehabilitation holds on motorized roads** Lokshin and Georgia (rural) Road rehabilitation No effect on agricultural pro- 17%↓ in time for Yemtsov (2005) duction or employment ambulance to arrive Dercon et al. Ethiopia (rural) Access to all- 7%↓ in poverty** (2008) weather road 16%↑ in consumption** Jacoby and Madagascar Natural variation in 50%↑ in income road elimi- Minten (2009) (rural) transport costs nated transport costs** Khandker et al. Bangladesh Road improvement 5%↓ in fertilizer prices 8-38%↑ in girls’ school (2009) (rural) 3%↑ in agricultural wage** attendance* 8-10%↑ in consumption** No effect on boys’ school attendance 90 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Danida (2010) Nicaragua Road improvement 17%↑ in employment*** 5.8 min/km↓ to health (rural) 16%↑ in within-community care post*** employment*** No effect on land value Cuong (2011) Vietnam (rural) Access to all- 9%↑ in income** No effect on school weather road attendance Khandker et al. Bangladesh Road improvement 10%↑ in household No effect on school (2011) (rural) expenditure** attendance 14%↑ in nonagricultural working days** Mu and van de Vietnam (rural) Road rehabilitation 9%↑ in market availability* 17-30%↑ in primary Walle (2011) and construction school completion** No effect on secondary school attendance Bell and van Del- Orissa, India All-weather road 5%↑ in output prices** 67%↓ in school No effect on mortality len (2012) (rural) construction absences** Rand (2012) Nicaragua Road improvement 31%↑ in hours worked*** (rural) Iimi and Lancelot Tocantins, Road and bridge 37%↓ in travel time to city 59%↑ in girls’ school No effect on health (forthcoming) Brazil (rural) improvement center** attendance*** condition 8%↑ in public bus use** 4%↑ in vehicle ownership* 17%↑ in household income* Iimi and East African Natural variation in 5%↑ in coffee production in Humphrey community rural accessibility Kenya and Tanzania** (forthcoming) (regional) 2–4%↑ in maize production in Tanzania** Iimi and Liberia Natural variation in 56.4%↑ in number of Kulwinder (2018) (national) road density patients who visited a health facility * 91 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Appendix B. Set-Up for Road Network Evaluation Tools Analysis 163.  This appendix presents the underlying set-up 165.  RONET calculates, for all standards applied to a for the Road Network Evaluation Tools (RONET) anal- road class, the corresponding stream of road agency ysis of the Mozambique road network. RONET is an costs (rehabilitation and maintenance), road user costs, Excel-based model for assessing the performance and total transport costs (sum of road agency and user of road maintenance and rehabilitation policies and costs) over the evaluation period; computes the pres- the importance of the road sector for the economy. It ent value of the total transport costs for each standard assesses current network conditions and computes the at the given input discount rate; and determines the asset value of the network and road network moni- optimal standard for the road class, which is the one toring indicators. It uses country-specific relationships that yields the lowest present value of total transport between maintenance spending and road conditions costs. Once the performance of each road class is eval- and between road conditions and road user costs uated under different standards, RONET presents the to assess the performance over time of the network corresponding results for a series of budget scenar- under different road project standards. It determines ios. The optimal scenario applies the optimal standard the proper allocation of expenditures among recurrent of each road class on all road classes in the network. maintenance, periodic maintenance, and rehabilitation Once RONET determines the optimal scenario, it eval- road projects. Finally, RONET determines the funding uates other budget scenarios with lower road agency gap, defined as the difference between current pres- expenditures than needed for the optimal scenario. ervation project spending and required spending and the effect of underspending on transport costs. 164.  The RONET Performance Assessment module assesses the consequences of various budget scenar- Road Network Data ios that represent different levels of road preservation project expenditures over time. The consequences 166.  The evaluation was based on Mozambique’s road are presented on road project requirements, financial network conditions and traffic data as of 2015/16 as cost, road conditions, asset value, road user costs, and given on the HDM-4 (a software package and associ- total transport costs. This module evaluates first the ated documentation for analyzing, planning, managing, performance of each road class of the network under and appraising road maintenance, improvements, and different road project standards over a 20-year evalu- investment decisions) road network file that the Admin- ation period. Road class is characterized according to istração Nacional de Estradas (ANE) provided (network. functional classification, surface type, traffic, and condi- mdb). The road network totals 6,677 km of paved roads tion. The very high standard indicates a policy without subdivided into two, three, four, and six lanes and pri- budget constraints and with a high frequency of road mary, secondary, and tertiary roads (table B.1). projects, and the high, medium, low, and very low stan- dards indicate decreasing frequencies of road projects 167.  Table B.2 shows the network distribution accord- and corresponding annualized road project expendi- ing to condition. Roads were classified into the follow- tures. The “do minimum” standard indicates a policy ing road condition classes based on their roughness. under which the only capital road project conducted over the evaluation period is reconstruction at a very high roughness. 92 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table B.1. Road Network on HDM-4 File According to Table B.2. Road Condition Classes, Based on Interna- Number of Lanes tional Roughness Index Road Number of lanes Total Condition class International Roughness Index, m/km classification 2 3 4 6 Very good < 2.5 Kilometers Good 2.5–3.5 Primary or trunk 4,283 427 38 12 4,760 Fair 3.5–5.5 Secondary or main 1,034 10 28 1,072 Poor 5.5–10.5 Tertiary or local 836 8 845 Very poor > 10.5 Total 6,154 446 66 12 6,677 Percentage The road network was in generally good condition, with 90 percent in very good, good, or fair condition. Primary or trunk 90 9 1 0 100 Secondary or main 96 1 3 0 100 Tertiary or local 99 1 0 0 100 Total 92 7 1 0 100 Table B.3. Road Network on HDM-4 File According to International Roughness Index Road classification International Roughness Index Total <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Kilometers Primary or trunk 974 2,115 1,454 216 0 4,760 Secondary or main 104 394 446 128 0 1,072 Tertiary or local 11 261 280 293 0 845 Total 1,089 2,770 2,180 638 0 6,677 Percentage Primary or trunk 20 44 31 5 0 100 Secondary or main 10 37 42 12 0 100 Tertiary or local 1 31 33 35 0 100 Total 16 41 33 10 0 100 93 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table B.4. Road Network on HDM-4 File According to Daily Traffic Road classification Average annual daily traffic (number of vehicles) Total <300 300–1,000 1,000–3,000 3,000–10,000 >10,000 Kilometers Primary or trunk 704 2,178 1,253 382 242 4,760 Secondary or main 192 789 91 1,072 Tertiary or local 548 164 124 9 845 Total 1,444 3,132 1,467 391 242 6,677 Percentage Primary or trunk 15 46 26 8 5 100 Secondary or main 18 74 9 0 0 100 Tertiary or local 65 19 15 1 0 100 Total 22 47 22 6 4 100 168.  Table B.4 shows network distribution according effect of weather events such as cyclones, which to traffic, represented by annual average daily traffic affected the condition of the road network. Therefore, in number of vehicles. Forty-seven percent of the road the estimated 2020 road network condition is the best network carries300 to 1,000 vehicles per day, and 91 approximation that can be made with the available percent carries fewer than 3,000. data but might not represent the actual condition of the network. Nevertheless, is sufficient for a macro 169.  The data provided on the HDM-4 file is from evaluation of the road network, such as the one done 2015/16, so for the RONET evaluation, available rough- using RONET. ness and traffic data were increased to represent 2020 conditions. The roughness data were increased by 2.9 170.  The traffic was increased by 4.9 percent per year percent per year based on the roughness progres- based on annual traffic growth defined in the HDM-4 sion that the HDM-4 model gave for Mozambique’s model for Mozambique’s roads,33 which the gross roads.32 Table B.5 shows the estimated condition of domestic product (GDP) increase of 3.7 percent per year the network in 2020 based on that roughness increase from 2015 to 2020 corroborates. Table B.6 shows the assumption. It was estimated that 83 percent of the estimated traffic of the network in 2020 based on that road network would be in very good, good, or fair traffic increase assumption. It was estimated that 48 condition in 2020. The evaluation did not consider road percent of the road network would carry 300 to 1,000 projects completed (periodic maintenance and reha- vehicles per day in 2020 and that 88 percent of the net- bilitation projects) from 2015 to 2020 or the negative work would carry fewer than 3,000 vehicles per day. 32 The road network links defined in 2015 were evaluated using HDM-4 to obtain the predicted roughness progression from 2015 to 2020 under routine main- tenance activities. The resulting average roughness progression found was 2.9 percent per year. 33 The Mozambique HDM-4 model configuration defines an elasticity of the annual traffic growth rate with per capita gross domestic product of 1.5. 94 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table B.5. Simulated 2020 Road Network According to International Roughness Index Road classification International Roughness Index Total <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Kilometers Primary or trunk 458 1,730 2,106 416 50 4,760 Secondary or main 54 130 572 316 1,072 Tertiary or local 176 283 386 845 Total 513 2,036 2,961 1,118 50 6,677 Percentage Primary or trunk 10 36 44 9 1 100 Secondary or main 5 12 53 29 0 100 Tertiary or local 0 21 33 46 0 100 Total 8 30 44 17 1 100 Table B.6. Simulated 2020 Daily Traffic Road classification Average annual daily traffic (number of vehicles) Total <300 300–1,000 1,000–3,000 3,000–10,000 >10,000 Kilometers Primary or trunk 403 2,126 1,463 491 277 4,760 Secondary or main 88 893 91 1,072 Tertiary or local 543 154 114 34 845 Total 1,034 3,173 1,668 525 277 6,677 Percentage Primary or trunk 8 45 31 10 6 100 Secondary or main 8 83 9 0 0 100 Tertiary or local 64 18 13 4 0 100 Total 15 48 25 8 4 100  The road network data provided in the HDM-4 road network file are for two-, three-, four-, and six-lane 171. roads, but for a RONET evaluation, the road network must be converted to a two-lane-equivalent road network. For example, 10 km of a four-lane road is equivalent to 20 km of a two-lane-equivalent road with half the traffic. Table B.7 shows that the road network consisted of 6,989 km of two-lane-equivalent roads, which were evaluated using RONET, and 5,034 km of which were primary roads (72 percent) and 1,106 km secondary roads (16 percent). 95 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Table B.7. Road Network Two-Lane-Equivalent Length  The RONET evaluation used the two-lane-equiv- 172. Road classification Actual length Two-lane- equivalent length alent network length distribution per road type (pri- mary, secondary, tertiary), traffic class (five traffic Kilometers range function of average annual daily traffic in num- Primary or trunk 4,760 5,034 ber of vehicles per day), and condition class (very good, Secondary or main 1,072 1,106 good, fair, poor, very poor) function-of-roughness ranges. Table B.8 shows the distribution of the net- Tertiary or local 845 849 work according to network type, traffic, and condition. Total 6,677 6,989 Table B.8. Distribution of Network According to Network Type, Traffic, and Condition Road classification Daily number of International Roughness Index (m/km) Total vehicles <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Kilometers Primary or trunk < 300 10 234 185 429 300–1,000 87 598 1,157 334 50 2,226 1,000–3,000 163 710 530 67 1,469 3,000–10,000 185 203 163 17 566 > 10,000 69 122 153 343 Total 513 1,867 2,186 417 50 5,034 Secondary or main < 300 12 76 88 300–1,000 54 130 546 196 927 1,000–3,000 42 49 91 3,000–10,000 > 10,000 Total 54 130 600 321 0 1,106 Tertiary or local < 300 100 203 242 545 300–1,000 32 65 58 154 1,000–3,000 35 17 64 116 3,000–10,000 9 25 34 > 10,000 Total 54 130 600 321 0 1,106 Total < 300 10 334 400 318 0 1,062 300–1,000 142 760 1,767 588 50 3,307 1,000–3,000 163 745 589 179 0 1,676 3,000–10,000 185 212 163 41 0 601 > 10,000 69 122 153 0 0 343 Total 568 2,173 3,072 1,126 50 6,989 96 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Road classification Daily number of International Roughness Index (m/km) Total vehicles <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Percentage Primary or trunk < 300 2 55 43 0 0 100 300–1,000 4 27 52 15 2 100 1,000–3,000 11 48 36 5 0 100 3,000–10,000 33 36 29 3 0 100 > 10,000 20 36 44 0 0 100 Total 10 37 43 8 1 100 Secondary or mail < 300 0 0 14 86 0 100 300–1,000 6 14 59 21 0 100 1,000–3,000 0 0 46 54 0 100 3,000–10,000 > 10,000 Total 5 12 54 29 0 100 Tertiary or local < 300 0 18 37 44 0 100 300–1,000 0 21 42 38 0 100 1,000–3,000 0 30 15 55 0 100 3,000–10,000 0 28 0 72 0 100 > 10,000 Total 0 21 34 46 0 100 Total < 300 1 31 38 30 0 100 300–1,000 4 23 53 18 2 100 1,000–3,000 10 44 35 11 0 100 3,000–10,000 31 35 27 7 0 100 > 10,000 20 36 44 0 0 100 Total 8 31 44 16 1 100 97 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING  Table B.9 shows the estimated 2020 road network distribution according to network type and condition. 173. Table B.9. Estimated 2020 Road Network Distribution According to Network Type and Condition Road classification International Roughness Index Total <2.5 2.5–3.5 3.5–5.5 5.5–10.5 >10.5 Kilometers Primary or trunk 513 1,867 2,186 417 50 5,034 Secondary or main 54 130 600 321 0 1,106 Tertiary or local 0 176 285 388 0 849 Total 568 2,173 3,072 1,126 50 6,989 Percentage Primary or trunk 10 37 43 8 1 100 Secondary or main 5 12 54 29 0 100 Tertiary or local 0 21 34 46 0 100 Total 8 31 44 16 1 100  Table B.10 shows the estimated 2020 road network distribution according to network type and traffic. 174. Table B.10. Estimated 2020 Road Network Distribution According to Network Type and Traffic Road classification Average annual daily traffic (number of vehicles) Total <300 300–1,000 1,000–3,000 3,000–10,000 >10,000 Kilometers Primary or trunk 429 2,226 1,469 566 343 5,034 Secondary or main 88 927 91 0 0 1,106 Tertiary or local 545 154 116 34 0 849 Total 1,062 3,307 1,676 601 343 6,989 Percentage Primary or trunk 9 44 29 11 7 100 Secondary or main 8 84 8 0 0 100 Tertiary or local 64 18 14 4 0 100 Total 15 47 24 9 5 100 98 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING  It was estimated that the road network would carry 6,016 million vehicle-km per year in 2020, of which 91 175. percent would travel on primary roads that account for 72 percent of the network length. The average traffic would be approximately 2,358 vehicles per day, of which 2,983 would use primary roads and 726 would use sec- ondary roads (table B.11). Table B.11. Road Network Condition and Traffic Road classification Length (km) International Good or fair condi- Annual average daily traffic Use (million Roughness Index tion (%) (number of vehicles) vehicle-km) Primary or trunk 5,034 4.1 91 2,983 5,481 Secondary or main 1,106 5.2 71 726 293 Tertiary or local 849 5.8 54 780 242 Total 6,989 4.4 83 2,358 6,016 Unit Road Preservation Project Costs 176.  Table B.12 presents estimated average unit costs of capital projects in U.S. dollars per kilometer for a two- lane road for different road classes. The capital projects to be done on a road are a function of current road conditions. According to the RONET logic, roads in good or fair condition (sustainable network) require periodic maintenance , and roads in poor or very poor condition require rehabilitation . The capital unit costs vary accord- ing to road network type because of their different standards. The estimated costs given in the table are based on data provided in the Mozambique ANE 2019 report. Table B.12. Unit Costs of Asphalt Concrete Projects Road classification Road work Road work type Primary or trunk Secondary or main Tertiary or local classification U.S. dollars/km Good Periodic Preventive treatment 29,250 27,000 22,500 maintenance Fair Resurfacing (5-cm overlay) 128,050 118,200 98,500 Poor Rehabilitation Strengthening (10-cm overlay) 256,100 236,400 197,000 Very poor Reconstruction 360,000 288,000 210,000  Routine maintenance costs were estimated to be USD2,000 per km per year based on an average of past 177. ANE expenditures and road projects for routine maintenance activities on paved roads from 2015 to 2018. RONET was not used to optimize routine maintenance project requirements of the road network. RONET optimizes only capital road projects. 99 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Vehicle Fleet Characteristics 178.  Table B.13 shows the 2020 vehicle fleet characteristics and economic unit costs used in the RONET evalua- tion. Economic costs are financial costs (market costs) net of taxes and subsidies. The estimated data given below are based on the data provided in the Mozambique ANE 2019 report. Table B.13. 2020 Vehicle Fleet Characteristics and Economic Unit Costs Used in Road Network Evaluation Tools Evaluation Name Economic or financial unit costs New New tire Fuel (USD/ Lubricating Main- Crew wages Overhead Passenger Passenger vehicle (USD/ liter) oil (USD/ tenance (USD/h) (USD/y) work time nonwork time (USD/ tire) liter) labor (USD/h) (USD/h) vehicle) (USD/h) Car medium 19,768 53 0.80 4.24 4.40 0.00 150 1.45 0.30 Delivery vehicle 28,170 69 0.80 4.24 4.40 1.25 160 1.45 0.30 Truck light 21,060 145 0.76 4.24 4.40 1.25 350 0.50 0.15 Truck medium 40,939 145 0.76 4.24 4.40 1.25 750 0.50 0.15 Truck heavy 95,250 198 0.76 4.24 4.40 1.25 1,480 0.50 0.15 Truck articulated 116,068 198 0.76 4.24 4.40 1.25 2,310 0.50 0.15 Bus medium 45,455 92 0.76 4.24 4.40 1.25 350 0.50 0.15 Bus heavy 76,590 198 0.76 4.24 4.40 1.25 730 0.50 0.15 Basic vehicle fleet characteristics Annual Annual Service life Private use Number Work Gross Annual Cargo time km working (years) (%) of passengers vehicle interest (USD/h) driven hours passengers trips (%) weight (%) (tons) Car medium 23,000 575 10 100 3 75 1.20 6 0.00 Delivery vehicle 23,000 575 10 0 3 75 1.80 6 0.00 Truck light 23,000 1,500 10 0 1 0 2.00 6 2.18 Truck medium 30,000 1,800 12 0 1 0 12.00 6 2.18 Truck heavy 40,000 2,500 12 0 1 0 15.00 6 2.18 Truck articulated 80,000 2,900 12 0 1 0 28.00 6 2.18 Bus medium 32,000 2,000 10 0 15 75 5.00 6 0.00 Bus heavy 70,000 2,300 12 0 50 75 12.00 6 0.00 100 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING References ADB (Asian Development Bank). 2018. Guide to Performance-Based Road Maintenance Contracts. Mandaluyong, Philippines: ADB. AfDB (African Development Bank). 2014. Study on Road Infrastructure Costs: Analysis of Unit Costs and Cost Over- runs of Road Infrastructure Projects in Africa. Abidjan, Côte d’Ivoire: AfDB. ANE (Administracao Nacional de Estradas). 2014. Economic and Social Plan/Integrated Road Sector Program (PES/PRISE): Report for the First Half of 2014. Mozambique: ANE. ———. 2019. Manual on Direct and Indirect Costs Related to Climate Adaptation Measures and Maintenance of Road Network. Mozambique: ANE. ———. 2020. Short Term Consultancy Services for Support to Public Private Partnership (PPP) Program Development of the Road Sector in Mozambique. Mozambique: ANE. Arrow, Kenneth, Theodore Harris, and Jacob Marschak. 1951. “Optimal inventory policy.” Econometrica 19(3): 205-272. Bajari, P., & Tadelis, S. 2001. “Incentives Versus Transaction Costs: A Theory of Procurement Contracts.” RAND Journal of Economics 32(3): 387–407. Beira Corridor. 2015. Beira Agricultural Growth Corridor Delivering the Potential. Bell, Clive and van Dillen, Susanne. 2012. “How Does India’s Rural Roads Program Affect the Grassroots? Findings from a Survey in Orissa.” World Bank Policy Research Working Paper No. 6167. World Bank, Washington, DC. Benmaamar, M. 2006. Financing of Road Maintenance in Sub-Saharan Africa, Discussion Paper no. 6, Road Management and Financing Series. Sub-Saharan Africa Transport Program. Benfica, Rui, Duncan Boughton, Bordalo Mouzinho, and Rafael Uaiene. 2014. “Food Crop Marketing and Agricultural Productivity in a High Price Environment: Evidence and Implications for Mozambique.” Research Paper 76E. Ministry of Agriculture, Maputo, Mozambique. Benson, Todd, Tewodaj Mogues, and Sileshi Woldeyohannes. 2014. “Assessing Progress Made Toward Shared Agricultural Transformation Objectives in Mozambique.” IFPRI Discussion Paper 01370. Blanc-Brude, F., Goldsmith, H., & Välilä, T. 2006. “Ex Ante Construction Costs in the European Road Sector: A Comparison of Public-Private Partnerships and Traditional Public Procurement.” Economic and Financial Report 2006/01, European Investment Bank. Bourke, Lisa, John Humphreys, John Wakerman, and Judy Taylor. 2012. “Understanding Rural and Remote Health: A Framework for Analysis in Australia.” Health and Place 18: 496–503. Bravo-Ortega, and Lederman. 2004. “Agricultural Productivity and Its Determinants: Revisiting Interna- tional Experiences.” Estudios de Economia 31(2): 133-163. Briceno-Garmendia et al. 2008. “Financing Public Infrastructure in Sub-Saharan Africa: Patterns and Emerg- ing Issues.” AICD Background paper 15. Cohan, Harry, and Frank Southworth. 1999. “On the Measurement and Valuation of Travel Time Variability Due to Incidents on Freeways.” Journal of Transportation and Statistics 2 (2): 123–131. 101 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Cuong, N. V. (2011). “Estimation of the impact of rural roads on household welfare in Viet Nam.” Asia-Pacific Development Journal 18(2): 105–135. Danida. 2010. Impact evaluation of Danida support to rural transport infrastructure in Nicaragua. Dercon, Stefan, Daniel Gilligan, John Hoddinott and Tassew Woldehanna. 2008. “The Impact of Agricultural Extension and Roads on Poverty and Consumption Growth in Fifteen Ethiopian Villages.” American Jour- nal of Agricultural Economics 91(4): 1007-1021. DHS (Demographic and Health Surveys). 2019. Mozambique MIS, 2018—MIS Final Report. Rockville, MD: INS e ICF. Escobal, J., and Ponce, C. 2002, ‘The Benefits of Rural Roads: Enhancing Income Opportunities for the Rural Poor.” GRADE, Working Paper 40–1, Lima. Espinet, Xavier, and Julie Rozenberg. 2018. “Prioritization of Climate Change Adaptation Interventions in a Road Network Combining Spatial Socio-Economic Data, Network Criticality Analysis, and Flood Risk Assessments.” Transportation Research Record 2672 (2): 44-53. Espinet, Xavier, et al. 2018. “Piloting the Use of Network Analysis and Decision-Making Under Uncertainty in Transport Operations : Preparation and Appraisal of a Rural Roads Project in Mozambique Under Changing Flood Risk and Other Deep Uncertainties.” Policy Research Working Paper 8490. World Bank, Washington, DC. Fujita, Masahisa, Paul Krugman, and Anthony Venables. 1999. The Spatial Economy. MIT Press. Fundo de Estradas. 2018. “About Us.” http://fe.gov.mz/index.php/sobre-fe/quem-somos/historial Gamm, Larry, Graciela Castillo, and Stephanie Pittman. 2003. “Access to Quality Health Services in Rural Areas: Primary Care: A Literature Review.” In Rural Healthy People 2010, edited by Larry Gamm, Linnae Hutchison, Betty Dabney, and Alicia Dorsey. http://sph.tamhsc.edu/centers/rhp2010/publications.htm Gonzalez-Navarro, Marco, and Climent Quintana-Domeque. 2016. “Paving Streets for the Poor: Experimen- tal Analysis of Infrastructure Effects.” The Review of Economics and Statistics 98(2): 254-267. Governo de Moçambique. 2015. “Sector de estradas pode rever taxa de combustível.” https://www.portaldogoverno.gov.mz/por/Imprensa/Noticias/Sector-de-estradas-pode-rever-taxa-de-combustivel. Gradín, Carlos, and Finn Tarp. 2019. “Investigating Growing Inequality in Mozambique.” South African Jour- nal of Economics 87: 110-138. Halpern, Benjamin, et al. (2015). “Spatial and Temporal Changes in Cumulative Human Impacts on the World’s Ocean.” Nature Communications 6: 7615. Holl, Adelheid. 2004. “Manufacturing Location and Impacts of Road Transport Infrastructure: Empirical Evi- dence from Spain.” Regional Science and Urban Economics 34, no. 3: 341-63. Hussein, Azhar, and Finorio Castigo. 2012. District Poverty Maps for Mozambique: 1997 and 2007: Based on Consumption Adjusted for Calorie Underreporting. Discussion Papers No. 74E. Maputo, Mozambique: National Directorate of Studies and Policy Analysis, Ministry of Planning and Development. IESE (Instituto de Estudos Sociais e Economicos). 2016. “Introduction to the Public Debt Problematic: Con- text and Immediate Questions.” IDeIAS 85e. 102 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Iimi, Atsushi, and Kulwinder Rao. 2018. “Transport Connectivity and Health Care Access: Evidence from Liberia.” Policy Research Working Paper No. 8413. World Bank, Washington, DC. Iimi, Atsushi. 2020. “Performance-Based Road Contracts in Zambia.” Review of Industrial Organization 57: 107–129. Iimi, Atsushi. 2021a. Agriculture Production and Transport Connectivity: Evidence from Mozambique. Policy Research Working Paper forthcoming. World Bank, Washington, DC. Iimi, Atsushi. 2021b. Transport Connectivity and Firm Productivity: Evidence from Mozambique. Policy Research Working Paper forthcoming. World Bank, Washington, DC. Iimi, Atsushi. 2021c. Estimating the Impact of Improved Roads on Healthcare Access in Mozambique. Policy Research Working Paper forthcoming. World Bank, Washington, DC. IMF (International Monetary Fund). 2013. Public Finance Management and Its Emerging Architecture. Washington, DC: International Monetary Fund. ———. 2019. “Republic of Mozambique: 2019 Article IV Consultation: Press Release, Staff Report, and State- ment by the Executive Director for the Republic of Mozambique.” IMF Country Report No. 19/166. IMF, Washington, DC. ———. 2019a. “Mozambique 2019 Article IV Consultation.” IMF Country Report No. 19/166. IMF, Washing- ton, DC INE (Instituto Nacional de Estatistica). 2015. Statistical Yearbook 2013. Maputo, Mozambique: INE. Institute of Medicine. 2005. “Quality Through Collaboration: The Future of Rural Health.” http://www.iom. edu/Reports/2004/Quality-Through-Collaboration-The-Future-of-Rural-Health.aspx#sthash.nqB3dnEa.dpuf Jacoby, H. G., and Minten, B. 2009. “On Measuring the Benefits of Lower Transport Costs.” Journal of Devel- opment Economics 89(1): 28–38. Khandker, Shahidur, Zaid Bakht, and Gayatri Koolwal. 2009. “The Poverty Impact of Rural Roads: Evidence from Bangladesh.” Economic Development and Cultural Change, 57(4): 685-722. Krugman, Paul. 1991. “Increasing Returns and Economic Geography.” Journal of Political Economy 99(3): 483-499. Laditka, James, Sarah Laditka, and Janice Probst. 2009. “Health Care Access in Rural Areas: Evidence that Hospitalization for Ambulatory Care-Sensitive Conditions in the United States May Increase with the Level of Rurality.” Health and Place 15: 761–70. Lancelot, Eric. 2010. “Performance Based Contracts in the Road Sector: Towards Improved Efficiency in the Management of Maintenance and Rehabilitation.” Transport Papers TP-31, The World Bank Group. Lee, Ki-Dong, Seok-Joon Hwang, and Min-hwan Lee. 2012. “Agglomeration Economies and Location Choice of Korean Manufactures Within the United States.” Applied Economics 44: 189-200. Li, Han, and Zhigang Li. 2013 “Road Investments and Inventory Reduction: Firm Level Evidence from China.” Journal of Urban Economics 76: 43-52. 103 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Limão, Nuno, and Anthony Venables. 2001. “Infrastructure, Geographical Disadvantage, Transport Costs, and Trade.” The World Bank Economic Review 15, no. 3: 451-79. Lishner, Denise, Roger Rosenblatt, Laura-Mae Baldwin, and Gary Hart. 2000. “Emergency Department Use by the Rural Elderly.” Journal of Emergency Medicine 18 (3): 289–97. Lokshin, M., and Yemtsov, R. 2005. “Has Rural Infrastructure Rehabilitation in Georgia Helped the Poor?” The World Bank Economic Review, 19(2): 311–333. Luis, Antonio dos Anjos, and Pedro Cabral. 2016. “Geographic Accessibility to Primary Healthcare Centers in Mozambique.” Equity in Health 15: 173. Maina, J., Ouma, P.O., Macharia, P.M. et al. 2019. “A Spatial Database of Health Facilities Managed by the Public Health Sector in Sub Saharan Africa.” Scientific Data 6 (134). https://doi.org/10.1038/s41597-019-0142-2. Mare, David, and Daniel Graham. 2013. “Agglomeration Elasticities and Firm Heterogeneity.” Journal of Urban Eco- nomics 75: 44-56. McFadden, Daniel. 1974. “Conditional Logit Analysis of Qualitative Choice Behavior.” In Frontiers in Economet- rics, edt. by P. Zarembka, pp. 105-142. Academic Press. Ministry of Agriculture. 2010. Strategic Plan for Agricultural Development PEDSA 2010-2019: For an Integrated, Prosperous, Competitive and Sustainable Agriculture Sector. Maputo, Mozambique: Ministry of Agriculture. Ministry of Industry and Trade. n.d. Investment Opportunities in the Industrial Sector. National Directory of Industry, Ministry of Industry and Trade, Mozambique. Ministry of Planning and Development. 2013. 2012 Survey of Mozambican Manufacturing Firms. Maputo, Mozambique: National Directorate of Studies and Policy Analysis, Ministry of Planning and Development. Mu, Ren, and Dominique van de Walle. 2011 “Rural Roads and Local Market Development in Vietnam.” The Journal of Development Studies, 47(5): 709-734. Pauw, Karl, James Thurlow, Rafael Uaiene, and John Mazunda. 2012. “Agricultural Growth and Poverty in Mozambique.” Mozambique Strategy Support Program Working Paper 2. Roberts, Peter, KC Shyam and Cordula Rastogi. 2006. “Rural Access Index: A Key Development Indicator.” Transport Papers TP-10. The World Bank Group. PPIAF (Public-Private Infrastructure Advisory Facility). 2009. Toolkit for Public-Private Partnerships in Roads and Highways. ———. 2017. Toll-Road PPPs: Identifying, Mitigating and Managing Traffic Risk. Washington, DC: World Bank Group. Procher, Vivien. 2011. “Agglomeration Effects and the Location of FDI: Evidence from French First-Time Movers.” Annals of Regional Science 46: 295-312. Rand, J. 2011. “Evaluating the Employment-Generating Impact of Rural Roads in Nicaragua.” Journal of Devel- opment Effectiveness 3(1): 28–43. Rozenberg, Julie, et al. 2017. “Improving the Resilience of Peru’s Road Network to Climate Events.” Policy 104 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Research Working Paper 8013. World Bank, Washington, DC. Schwitters, Amee, Philip Lederer, Leah Zilversmit, Paula Samo Gudo, Isaias Ramiro, Luisa Cumba, Epifanio Mahagaja, and Kebba Jobarteh. 2015. “Barriers to Health Care in Rural Mozambique: A Rapid Ethno- graphic Assessment of Planned Mobile Health Clinics for ART.” Global Health: Science and Practice 3 (1). Simler, Kenneth, and Virgulino Nhate. 2005. “Poverty, Inequality and Geographic Targeting: Evidence from Small-Area Estimates in Mozambique.” FCND Discussion Paper 192. International Food Policy Research Institute, Washington, DC. Small, Martin, and Justin Runji. 2014. “Managing Road Safety in Africa: A Framework for National Lead Agencies.” SSATP Working Paper 101. Africa Transport Policy Programme, Washington, DC. Shirley, Chad, and Clifford Winston. 2004. “Firm Inventory Behavior and the Returns from Highway Infra- structure Investment.” Journal of Urban Economics 55, no. 2: 398-415. UNDP (United Nations Development Programme). 2019 Mozambique Cyclone Idai: Post-Disaster Needs Assess- ment. New York: UNDP. USAID (United States Agency for International Development). 2018. President’s Malaria Initiative: Mozambique Malaria Operational Plan FY18. Washington, DC: USAID. USAID (United States Agency for International Development). 2020. Global health – Mozambique. Available at https://www.usaid.gov/mozambique/global-health Warren-Rodríguez, Alex. 2010. “Uncovering Trends in the Accumulation of Technological Capabilities and Skills in the Mozambican Manufacturing Sector.” Oxford Development Studies 38 (2): 171–198. World Bank. 2016. Measuring Rural Access: Using New Technologies. Washington, DC: World Bank. ———. 2017. Enhancing the Climate Resilience of Africa’s Infrastructure: The Road and Bridge Sector. Washington, DC: World Bank. ———. 2018. “PPP Benchmark Procurement Report.” https://ppp.worldbank.org/ public-private-partnership/library/procuring-infrastructure-ppps-2018 ———. 2018. Enterprise Surveys: Mozambique: 2018 Country Profile. Washington, DC: World Bank. ———. 2019a. Mind the Rural Investment Gap: Disparities in Access to Basic Infrastructure and Directions for Mozambique’s Public Investment Program. Washington, DC: World Bank. ———. 2019b. Scaling Up Private Sector Participation in Road Asset Management in Sub-Saharan Africa. Washing- ton, DC: World Bank Group. ———. 2020a. Mozambique Country Private Sector Diagnostic. Washington, DC: World Bank Group. ———. 2020b. Strengthening Economic Management for Inclusive Growth: Annual Report July 2019 – June 2020. Washington, DC: World Bank Group. ———. 2020c. Mozambique Country Economic Memorandum: Reigniting Growth for All. Washington, DC: World Bank Group. ———. 2020d. Performance and Learning Review of the Country Partnership Strategy for the Republic of 105 INVESTING IN MOZAMBIQUE’S CONNECTIVITY: TOWARD EFFICIENT, SUSTAINABLE ROAD SPENDING Mozambique for the Period of FY17-FY21. Washington, DC: World Bank. ———. 2020e. Mozambique: Strengthening Economic Management for Inclusive Growth Annual Report: 2019/20. Washington, DC: World Bank Group. ———. 2021. World Development Indicators. Washington, DC: World Bank Group. ———. forthcoming. Mozambique Jobs Strategy Policy Note. Washington, DC: World Bank Group. Yusuf, Shahid, Kaoru Nabeshima, and Shoichi Yamashita (editors). 2008. Growing Industrial Clusters in Asia, World Bank: Washington, D.C. Zavale, Helder, Greenwell Matchaya, Delfim Vilissa, Charles Nhemachena, Sibusiso Nhlengethwa, and Duque Wilson. 2020. “Dynamics of the Fertilizer Value Chain in Mozambique.” Sustainability 12(11): 4691. https://doi.org/10.3390/su12114691.